A theoretical stochastic control framework for adapting radiotherapy to hypoxia

Area of Science:

• Radiation Oncology
• Medical Physics
• Computational Biology

Background:

• Hypoxia (insufficient oxygen) reduces tumor radiosensitivity, particularly in head and neck cancers, negatively impacting radiotherapy outcomes.
• Oxygen levels in tumors vary spatially and temporally, presenting a challenge for conventional radiotherapy planning.
• Advances in functional imaging may enable adaptive radiotherapy, adjusting treatment based on real-time hypoxia evolution.

Purpose of the Study:

• To theoretically investigate the benefits of adaptive radiotherapy planning for head and neck cancer.
• To develop and evaluate an approximation algorithm for dynamic treatment adaptation in the presence of hypoxia.

Main Methods:

• A stochastic control framework was employed using computer simulations of hypoxia evolution in head and neck cancer models.
• A certainty equivalent control algorithm, approximating the intractable exact solution, was developed, involving sequential convex programming.
• Numerical experiments utilized vector autoregressive processes to simulate spatiotemporal hypoxia dynamics on test cases.

Main Results:

• The proposed dynamic planning approach demonstrated potential for substantial improvement in reducing residual tumor cell counts.
• Simulations provided insights into the conditions under which dynamic planning offers the greatest advantages.
• The method successfully handled dose-volume constraints using a constraint generation technique within convex optimization.

Conclusions:

• Adaptive radiotherapy planning, by dynamically adjusting to hypoxia, shows promise for enhancing treatment efficacy in head and neck cancers.
• The developed certainty equivalent control algorithm offers a computationally feasible approach to dynamic treatment optimization.
• Further research into radiobiological parameters is needed to confirm the applicability of the model to various tumor types.