3187 - Modeling Radiotherapy and Chemotherapy Responses Using Pancreatic Tumor Organoids

Purpose/Objective(s): Pancreatic cancer has a poor prognosis with heterogeneity in patient responses to standard of care chemo, radiation, or chemoradiation therapy (CRT). Patient-derived tumor organoids (PDTOs) have emerged as a promising platform for testing new therapies and patient-specific responses to treatment. In this study, we developed a mathematical modeling framework to characterize PDTO treatment dynamics and provide quantitative insight into the timing and interactions between therapies.

Materials/Methods: We developed novel mathematical forecasting frameworks based on ordinary differential equations and characteristics of tumor responses to model the therapeutic effects of chemo- and radiation-induced killing in PDTOs. PDTOs from three patients were treated with radiation (4 Gy, 8 Gy), chemotherapy (FOLFIRINOX), and combined regimens. Diameters of 20–40 organoids per patient were measured from brightfield images up to 9 days post-treatment and used for model calibration. Model performance was assessed by average normalized mean squared error, cross-dataset validation, and comparison with alternative models.

Results: The proposed models accurately captured the growth dynamics of PDTO following chemo, radiation, and CRT. The fitted parameters, i.e., killing pattern, response window, and peak killing timing, revealed significant heterogeneities in treatment responses across different PDTOs. Chemotherapy demonstrated a pronounced early effect, peaking around days 3-6 and lasting approximately 10 days, while radiotherapy exhibited a delayed but stronger cytotoxic effect emerging around day 4 post-irradiation and a significantly higher secondary peak near day 8. CRT integrates the strengths of both modalities, producing a prolonged and intensified response window; modeling results suggest that the radiation-induced killing effect may play a dominant role in the combined interaction. The model further demonstrated predictive capacity across varying radiotherapy dose regimens and treatment conditions for individual organoid samples. Additionally, extended growth modeling revealed that ignoring the internal compression effect (i.e., inner necrosis exceeding outer proliferation) in larger organoids (~300 µm in diameter) may lead to underestimation of carrying capacity and overestimation of treatment response by regular growth models.

Conclusion: Our modeling frameworks demonstrated high accuracy in characterizing the heterogeneous therapeutic responses of PDTOs, providing quantitative insight into the dynamic killing patterns for each therapy. The models further enabled treatment-specific response decomposition and patient-level forecasting, offering a valuable tool to evaluate treatment effects, optimize regimen design, and inform clinical trial strategies, ultimately advancing personalized oncology.