
Mathematical Modeling and Optimization of Tumor and Therapeutic Cell Delivery Incorporating Interstitial Fluid Flow
Please login to view abstract download link
Chimeric antigen receptor (CAR) T-cell therapy is an increasingly common immune therapy for treatment of tumors [1]. Primary neurological tumors still remain largely difficult to treat with systemic delivery of CARs, due to blood brain barrier prevention of uptake. To overcome this challenge, some CAR-T trials have adopted the strategy of intratumoral delivery, placing cells directly in the treatment site using a surgical catheter [1]. This method directly circumvents the issue of vascular delivery to the brain parenchyma, and achieves greater total delivery to the tumor. It is known that both tumor cells and therapeutic cancer cells are more likely to invade in directions of increased interstitial fluid flow. To measure this fluid flow, we developed a novel method, localized convolutional function regression, which enables simultaneously physics inversion of 4D data through application of convolutional weak-derivative kernels and localized sparse regression of the results [2]. In a retrospective study of patients receiving CAR-T cells for high grade brain cancer, we investigated advection, perfusion, and diffusion as imaging biomarkers of prolonged survival. We found that patients with more heterogenous flow (p = 1.4E-3) and perfusion (p = 1.6E-3) profiles were statistically more likely to survive longer than those with more uniform flow and perfusion profiles. Further, we aim to utilize this retrospective data to develop a model of CAR-T cell delivery, where we can optimize the delivery location and therapeutic timing. This interstitial fluid flow and diffusion measured by LCFR applied to DCE-MRI imaging [2], and T-cell proliferation and exhaustion dynamics are informed by in-vitro experiments [2]. Using these preliminary methods on the largest 2D imaging slice, we are able to more than double the total population of CAR T-cells through spatial catheter placement optimization alone. We aim to further optimize the efficacy of CAR-T cell therapy by performing numerical experiments on digital twins receiving varied dose schedules. REFERENCES [1] Brown, Christine E., et al. "Locoregional delivery of IL-13Rα2-targeting CAR-T cells in recurrent high-grade glioma: a phase 1 trial." Nature medicine 30.4 (2024): 1001-1012. [2] Woodall, Ryan T., et al. "Model discovery approach enables noninvasive measurement of intra-tumoral fluid transport in dynamic MRI." APL bioengineering 8.2 (2024).