Computational prediction of tumor progression in brachytherapy

Permanent brachytherapy brain implants offer a method to deliver radiation therapy intraoperatively immediately following tumor resection. Compared to external beam radiation therapy, brachytherapy provides targeted, high-dose radiation with fewer side effects. GammaTile, an FDA-cleared device for brachytherapy, comprises four 131Cs radiation seeds embedded in a collagen tile, offering modular, localized radiation with biocompatible implant material. Studies have shown that GammaTile therapy improves overall survival for patients with recurrent glioblastoma (GBM). GammaTiles are placed at the surgical bed's areas with the highest risk of recurrence, as determined by neurosurgeons and radiation oncologists. The radiation dosage is calculated by a medical physicist using post-implant CT and MRI scans. However, the relationships between radiation dosage, tumor recurrence, and infiltration probabilities in GBM tumors remain largely unknown. Currently, no computational tools exist to predict how GBM cells might be affected by the radiation dose and migrate to distant sites. This proposal aims to simulate tumor progression in patients undergoing brachytherapy. In vitro 2D and 3D cell proliferation and migration (both random and directional) with localized radiation will be developed and studied. Previously biophysical models, such as the Cell Migration Simulator and Brownian Dynamics Model, will be further developed to computationally simulate and analyze the physical and molecular mechanisms of tumor cell growth and migration under localized radiation. Additionally, previously machine learning algorithms will be enhanced to extract single-cell features from simulation data and patient MRI images, and to predict tumor progression in GBM patients under brachytherapy. The computational tools developed will enable surgeons and oncologists to optimize patient-specific strategies, thereby improving patient outcomes in brachytherapy. Furthermore, these algorithms can be adapted to simulate tumor progression under drug and radiation interventions, potentially guiding the development of novel treatment strategies in cancer.