Self-reported race and ethnicity and genetic ancestry groups show distinct prevalence and outcomes across pediatric central nervous system tumors

Abstract BACKGROUND Race and ethnicity disparities have been identified in survival of pediatric central nervous system (CNS) tumors. No prior studies have investigated the comparative effects of self-identified race and ethnicity versus genetic ancestry on tumor prevalence and outcome. We examine these contributions in a retrospective analysis of pediatric patients with CNS tumors. METHODS Patients from Children’s Brain Tumor Network (CBTN; n=646) and Pacific Pediatric Neuro-Oncology Consortium (PNOC; n=35) with available clinical data and whole genome sequencing were included. Ancestry groups were predicted with somalier and genetic markers from reference individuals of known ancestry that participated in the 1000 Genomes Project: sub-Saharan African (AFR), admixed American (AMR), East Asian (EAS), European (EUR), and South Asian (SAS). Independence of predicted ancestry and self-reported race and ethnicity was assessed with chi-squared tests. Effects of predicted ancestry on survival were estimated using multivariate Cox models with molecular subtype and extent of tumor resection as covariates. RESULTS Predicted ancestry groups were highly concordant with self-reported race (χ2=897, p<0.001) and ethnicity (χ2=340, p<0.001). Patients from AFR ancestry were significantly enriched within ATRT (n=26; p=0.001) and neurofibroma plexiform (n=11; p=0.005) and patients with AMR ancestry were enriched within DMG (n=53; p<0.001). Patients diagnosed with H3 wildtype high-grade glioma (n=39) with non-EUR ancestry trended towards worse overall survival (OS; HR=2.0, p=0.07). Patients with ATRT (n=19) with AMR ancestry trended towards worse OS compared to AFRs (HR=4.1, p=0.09) and AFR patients with MB (n=78) trended toward worse OS compared to EUR patients (HR=3.7, p=0.07), including Group 4 MBs (n=39; HR=3.9, p=0.08). Patients with LGG (n=182) with non-EUR ancestry were more likely to have less than gross total/near total resection compared to patients with EUR ancestry (p=0.02). CONCLUSIONS Differences in ancestry are seen across pediatric CNS tumor entities and outcomes. Analyses are underway in an additional ~800 patients.