Objective We aim to investigate epidemiologic trends, treatment modalities, and overall survival in patients diagnosed with invasive spinal hemangioblastoma and to determine the predictors impacting overall survival and postoperative length of stay (LOS).

Methods We queried the National Cancer Database from 2004 to 2016 for patients diagnosed with invasive spinal hemangiomas. Demographic, clinical, and outcomes data were compiled. Five-year survival for each treatment was compared using the Kaplan-Meier (KM) and Nelson-Aalen (NA) curves with statistical comparisons based on the log-rank test. Multivariate analysis was performed to determine survival determinants. Supervised machine learning (ML) models were trained to predict survival and model performance was evaluated. Shapley additive explanations (SHAP) were calculated using the best-performing model for feature importance.

Results A total of 798 patients diagnosed with invasive spinal cord hemangioblastoma were analyzed. The most common treatment was surgical resection (N=687), followed by radiotherapy (N=67) and surgery combined with radiotherapy (N=50). Age (p<0.01), a Charlson comorbidity score (CDCC) of 2 or 3 (p=0.04; p=0.025), tumor size (p<0.01), and chemotherapy alone (p<0.01) are more likely to predict mortality. Surgical resection alone (p=0.02) was less likely to predict mortality. The best-performing ML model was Random Forest (99.9%). Increased odds of survival are associated with surgery alone or surgery with radiotherapy (p<0.05). SHAP showed that age, year of diagnosis, and days until radiation therapy has the highest impact on survival. CDCC of 2; Medicare or private insurance are more likely to predict postoperative LOS.

Conclusion This study provides the most comprehensive description of invasive hemangioblastomas’ epidemiologic, treatment, and survival trends. Our analysis demonstrates that surgical resection, either alone or coupled with radiotherapy, significantly improves long-term survival

Disclosures A. Ghaith: None. A. Bon Nieves: None. M. Ghanem: None. B. Bendok: None.