Predictive Risk Calculator of Racial and Socioeconomic Disparities of Overall 5-Year Survival in Treated Patients with Hemangioblastomas

Introduction: Hemangioblastomas are rare, benign tumors of the central nervous system. Understanding of socioeconomic factors, which can lead to disparities in diagnosis, access to care, treatment, and patient outcomes, remains limited for hemangioblastoma. This study aims to explore the socioeconomic disparities in hemangioblastoma and the associated outcomes.

Methods: Patients with hemangioblastoma were identified from the National Cancer Database (NCDB) between 2004 and 2017. The cohort was divided into three racial groups: White, Black, and Asian. Univariate analyses were conducted for the comparative study, Kaplan-Merier survival analysis was used to assess overall survival (OS), and multivariate Cox regression was performed for the risk factor study. Receiver operating characteristic (ROC) and C-index analyses were used to evaluate the predictive performance of machine learning models.

Results: A total of 716 patients were identified (White: 599, Black: 88, Asian: 29). Overall, the five-year mortality rate was highest in White patients (14.7%), followed by Black (5.7%) and Asian (3.4%) (p=0.019). The White group also had the highest age mean, followed by Black and Asian (p=0.009). Facility types were similar across races. However, patients accessed through the Comprehensive Community Cancer Program had the worst OS (p < 0.0001) and were associated with an increased risk (HR: 1.97, p=0.006). The White group had the highest proportion of patients covered by Medicare (21.5% p< 0.001), which was found to have the worst OS (p < 0.0001). The Asian and White groups dominated in patients in the highest income quartile, in contrast to the Black group with a noticeable proportion of patients in the lowest quartiles (p < 0.001). The black group also had a higher proportion of individuals with a Charlson-Deyo Comorbidity Score (CDCC) of 1 or above compared to the other two races (p=0.043). The Gradient Boosting Survival model was found to be the most predictive model, confirmed by ROC curves analysis (AUC=0.82) and C-index test (C-index=0.78), with age being the most influential prognostic factor.

Conclusion : This study reveals important socioeconomic disparities among hemangioblastoma patients, highlighting the influence of these factors on patient outcomes and underscoring the need for further research to address inequities. The link to the Gradient Boosting Survival model-based calculator: https://huggingface.co/spaces/AKG47/Spine_Tumors_race