Predicting Amplification of Using CpG Methylation Biomarkers in Neuroblastoma

Neuroblastoma is the most common extracranial solid tumor in childhood. Amplification of in neuroblastoma is a predictor of poor prognosis. DNA methylation data from the TARGET data matrix were stratified into amplified and non-amplified groups. Differential methylation analysis, clustering, recursive feature elimination (RFE), machine learning (ML), Cox regression analysis and Kaplan-Meier estimates were performed. 663 CpGs were differentially methylated between the two groups. A total of 25 CpGs were selected by RFE for clustering and ML, and a 100% clustering accuracy was obtained. ML validation on three external datasets produced high accuracy scores of 100%, 97% and 93%. Eight survival-associated CpGs were also identified. Therapeutic interventions may need to be targeted to patient subgroups.