Multivariable Normal Tissue Complication Probability Prediction of Early Grade Radiation-Induced Optic Neuropathy...

Area of Science:

• Radiation Oncology
• Ophthalmology
• Medical Physics

Background:

• Radiation-induced optic neuropathy (RION) affects approximately 4% of patients with extraocular tumors.
• Existing predictive models are limited, accurately identifying only irreversible, high-grade RION.

Purpose of the Study:

• To develop a sensitive multivariable Normal Tissue Complication Probability (NTCP) model.
• Integrate quantitative visual outcomes and dosiomics to predict early-grade RION (grade 2+).
• Focus on a cohort treated with pencil beam scanning proton therapy.

Main Methods:

• Prospective study of patients with head and neck or central nervous system tumors.
• Standardized ophthalmologic assessments and dosimetry analysis of optic pathways.
• Development of logistic regression and random forest models with feature selection.

Main Results:

• 23.5% of analyzed eyes developed grade 2+ RION within a median follow-up of 3.6 years.
• Random forest models with dosiomics achieved an area under the curve of 0.8, outperforming traditional models.
• Key predictors included baseline visual field deficit and optic chiasm V50 (volume receiving 50 Gy).

Conclusions:

• A 23.5% risk of grade 2+ RION was observed using visual field perimetry.
• Risk factors include gender, vascular comorbidities, pre-existing optic pathway damage, and optic chiasm dose.
• The developed NTCP model offers improved prediction of early-stage RION.