Intensity-modulated radiation therapy (IMRT) remains the standard-of-care radiation therapy
technique for most head-and-neck cancer (HNC) patients. Despite the relatively high survival
rate of HNC (65-75%), the survivors often suffer from post-treatment long-lasting side effects,
which negatively affect their quality of life. In an IMRT-based treatment plan, the daily doses
delivered to the tumor and organs-at-risk (OARs) remain fixed based on the premise that the
one-day inter-fraction suffices for each OAR’s recovery. However, a novel treatment planning
approach, referred to as temporally feathered radiation therapy (TFRT), proposes variable daily
dose delivery to the OARs, aiming at increasing the recovery time of OARs at the cost of
receiving higher doses once a week. While initial implementations of TFRT have shown
promising results in the clinic, they do not distinguish between the inherent radiobiological
differences among OARs for HNC patients, thus prescribing longer, yet uniform recovery time
for all OARs. By considering variable radiosensitivity profiles across multiple OARs for HNC
radiation therapy, we develop a mixed-integer nonlinear program (MINLP) that computes
specific recovery time for each OAR to minimize the toxicity burden to the patients. Our
proposed MINLP leverages the linear-quadratic model, which predicts each OAR’s survival
from radiation-induced damages. Our model is further flexible in addressing specific
prioritization of OARs based on the clinician's preferences. The result of our model indicates a
lower overall toxicity burden when OARs with higher radiosensitivity have longer recovery
times under the TFRT framework.

Short Bio