Normal Tissue Complication Probability Modeling for Proton Therapy.

Area of Science:

• Medical Physics
• Radiation Oncology
• Clinical Decision Support

Background:

• Accurate normal tissue complication probability (NTCP) models are essential for optimizing proton therapy planning and patient selection.
• Proton therapy's unique dose-response characteristics necessitate specialized toxicity models beyond those derived from photon therapy.

Purpose of the Study:

• To review the challenges in developing accurate proton-specific NTCP models.
• To examine the management of NTCP prediction uncertainty for reliable clinical decision-making in proton therapy.

Main Methods:

• Literature review of existing NTCP models and their applicability to proton therapy.
• Analysis of the endpoint-dependency of NTCP model accuracy.
• Assessment of clinical evidence for incorporating variable relative biological effectiveness.
• Evaluation of the attention paid to prediction uncertainty in clinical decision frameworks.

Main Results:

• Photon-derived NTCP models are not universally applicable to proton therapy, with accuracy being endpoint-dependent.
• Clinical evidence supporting the update of NTCP models with variable relative biological effectiveness is limited due to heterogeneity and inconsistent findings.
• The impact of NTCP prediction uncertainty on patient selection has been inadequately addressed.

Conclusions:

• Proton-specific NTCP models are required for certain toxicity endpoints.
• Further research is needed to develop a robust framework for NTCP-guided patient selection that accounts for both model accuracy and prediction confidence.
• Addressing model uncertainty is critical for reliable clinical decision-making in proton therapy.