Modeling Time-Dependent Recovery to Improve Esophagitis Prediction in Thoracic Reirradiation.

Area of Science:

• Radiation Oncology
• Medical Physics
• Clinical Oncology

Background:

• Re-irradiation requires balancing tumor control with adverse effects.
• Accurate toxicity prediction is crucial for safe re-irradiation.
• Time-dependent recovery models offer a potential improvement over current methods.

Purpose of the Study:

• To evaluate the feasibility of time-dependent recovery models for toxicity prediction in thoracic re-irradiation.
• To compare the predictive performance of different time-dependent models against conventional dose accumulation methods.

Main Methods:

• Sixty-five high-risk thoracic re-irradiation patients were analyzed.
• Dose accumulation was performed using voxel-wise EQD2 with deformable registration.
• Time-dependent models (mono-exponential, bi-exponential, reciprocal) were compared to direct accumulation using logistic regression and AUC analysis.

Main Results:

• The bi-exponential time-dependent recovery model demonstrated superior prediction of grade ≥ 2 esophagitis (AUC: 0.83).
• Time-dependent models significantly outperformed direct dose accumulation (AUC: 0.74).
• Patient history of esophagitis improved the bi-exponential model's performance.

Conclusions:

• Incorporating time-dependent recovery into dose accumulation is feasible and improves toxicity prediction accuracy.
• The bi-exponential model shows promise for optimizing re-irradiation doses.
• These findings support the development of advanced outcome models for personalized re-irradiation to reduce toxicity.