Impact of proton versus photon radiotherapy on overall survival in the management of spinal chordoma and mortality risk prediction: a nationwide analysis.

Purpose: This study aims to provide a national-level insight into the optimal management of spinal chordoma, a type of rare and complicated malignancy, and to identify effective radiation therapy (RT) strategies for patients with spinal chordoma by comparing photon versus proton therapies.

Methods: Patients with histologically confirmed spinal chordoma were identified from the NCDB. Patients' demographics, disease characteristics and treatments were compared between the radiation and no-radiation groups. Kaplan-Meier assessed 10-year overall survival (OS). Multivariate Cox regression and machine learning models-Random Survival Forest and Gradient Boosting-identified predictive factors and thresholds of tumor size for mortality risk, with SHAP used to quantify predictor importance.

Results: Of 1204 patients, 495 (41.1%) received RT; 73.9% received photons and 26.1% protons. GTR significantly improved OS compared to STR (p < 0.0001). RT was associated with reduced short-term morality at 90-days (p = 0.025) and 1-year (p = 0.016). Proton therapy showed superior 10-year OS compared to photon therapy (p = 0.019). Beam technology, BED dose, sequence, and reirradiation had minimal impact on OS. Cox regression identified age (HR = 1.03, p < 0.001), STR (HR = 1.42, p = 0.005), and proton therapy (HR = 0.58, p = 0.044) as independent mortality predictors. Machine learning identified tumor size thresholds (photons: 83 mm; protons: 70 mm) and ranked age, tumor size, and EOR as top survival predictors.

Conclusion: Aggressive surgical resection (GTR) remains the primary treatment for spinal chordoma. Proton therapy that allows for dose escalation provided 10-year survival benefits compared to photon therapy. Age, tumor size and EOR are important factors in predicting 10-year mortality risk.