SU-F-R-49: A Novel Kinetic Model for Prediction of Tumor Local Control for Patients with Lung Cancer

Purpose: Modeling tumor control probability (TCP) can help optimize treatment plans for better treatment outcomes. This study sought to quantify the radiobiological parameters of the TCP model for patients with lung cancer. Methods: A two-compartment kinetic model was developed to model tumor regression for five NSCLC patients. The model has three parameters: cell survival fraction, dead-cell-resolving time and tumor doubling time. The last one was extended to a function of tumor volume. Each of these patients was treated with 2 Gyx33 fractions. Daily CBCT images were acquired during the course of treatment. Gross tumor volume (GTV) was delineated on each of these CBCT images, and the visible tumor volumes were fitted to the kinetic model to optimize the parameters, where the Jacobian of this model was constrained by daily tumor volumetric changes. Results: Among the five patients, three had tumor recurrence: 455, 520 and 590 days after the completion of treatment. Recurrent tumor volumes were, respectively, 28.9, 13.7 and 4.8 cm3, measured at distant locations. With the assumption of cell density = 108 cells/cm3, tumor doubling times required for circulating tumor cells to progress to the recurrent volumes are 16.1, 19.2 and 23.1 days. By fitting the kinetic model with 33 measured tumor volumes, the average cell survival fraction for the five patients is 0.5±0.21, and tumor doubling times are in the range of 10∼35 days, comparable to the doubling time derived from tumor recurrence data. The doubling times modeled from invisible cells are different from the median (357 days) or mean (166 days) of the doubling time measured directly from visible tumor volumes. Conclusion: A kinetic model has been developed to simulate the process of tumor progressing from invisible cells to visible tumor volumes. This model may be useful for TCP prediction for patients with locally advanced lung cancer.