Altered Fraction During Radiation to Compensate for Development of Resistance and Intratumor Heterogeneity in Glioblastoma.

Purpose/Objective(s) Glioblastoma (GBM) is characterized with poor survival and local control outcomes despite attempts at dose escalation. Fractional equivalent dosing (FED) does not account for intratumor heterogeneity or development of resistance during treatment. We develop and validate a mathematical model to investigate the impact of ramping up radiation dose (RUD) during treatment. Materials/Methods A mathematical simulation of radiation treatment (mRT) was developed with the following parameters: 1. One billion starting cells 2. 72 hour doubling time 3. Intratumor heterogeneity: Ten distinct subpopulations for each patient that represent a normal distribution of their RSIab value (a previously described genomic model for estimating alpha and beta values based on gene expression) 4. Treatments simulated 5 days of RT with 2 day break 5. Cell death dictated by RSIab parameters and the LQ equation 1. LQ equation = SF = e^(-a*d-b*d^2) 6. Cells that survive are only allowed to double 7. If subpopulation reaches mRT output is surviving fraction at the end of the treatment. A simulation cohort of 40 GBM patients with available genomic information and RSIab was used to simulate the dynamics of FED (60 Gy in 30 fractions) vs. RUD (60 Gy in 30 fractions with an average of 2 Gy/fx qday increment of 0.1 Gy with a range of 0.5-3.5 Gy). A separate validation cohort derived from EORTC26951 high grade glioma trial with available RSIab and clinical endpoints were put into mRT for validation after locking down model parameters. The clinical end point was overall survival. Wilcoxon signed-rank test was used in the simulation cohort between FED vs RUD. In the validation cohort, univariate analysis of the SF for each patient was conducted using log-rank Kaplan-Meier curves with median dichotomization. Multivariate cox-regression analysis of SF was performed with SF and age as continuous variables and PCV as a categorical variable. Results The simulation cohort demonstrates a significant decrease in average alpha/beta ratio in the cohort with respective subpopulations during FED (a/b ratio of 7.95, 4.53, 3.07, and 1.608 after the 0th, 5th, 10th, and 30th fraction, respectively). This is contrast with RUD dosing that shows a less dramatic decrease as dose ramps up (a/b ratio of 7.95, 5.91, 3.93, and 1.94 for the 0th, 5th, 10th, and 30th fraction, respectively). The median SF at the end of treatment between FED (1%) vs. RUD (0.51%) after 60 Gy in 30 fractions had statistically significant difference (P = 0.02). In the validation cohort, the output of SF was statistically significant for overall survival on univariate (P = 0.001) and multivariate analysis (HR: 1.03-4.33 P = 0.041). Conclusion This work demonstrates RUD is superior to FED when considering intratumor heterogeneity in radiation treatment of Glioblastoma because it accounts for the selection of lower alpha/beta subpopulations. Author Disclosure A. Subramanian: None. C. Li: None. D.E. Oliver: None. J.G. Scott: None. J.F. Torres-Roca: None. J. Peacock: None.