DNA Methylation Profiling Classifies Neuroblastoma Into Three Clinically Distinct Subtypes

Neuroblastoma is one of the most clinically heterogeneous pediatric cancers, ranging from tumors that spontaneously regress to aggressive high-risk disease with mortality exceeding 50%. Current risk stratification relies on age, disease stage, MYCN amplification, and 11q deletion status. Recent evidence suggests telomere maintenance mechanisms (TMM) — either TERT reactivation or the alternative lengthening of telomeres (ALT) pathway — are powerful independent prognostic factors. This study evaluated whether a DNA methylation-based classifier could stratify neuroblastoma into clinically relevant molecular subgroups.

The investigators applied the Molecular Neuropathology (MNP) classifier v12.5 to two cohorts: a local Swedish cohort of 90 tumors and 213 publicly available samples from the US Therapeutically Applicable Research to Generate Effective Treatments (TARGET) database. Methylation arrays (Illumina 450K/EPIC) were used, and calibrated scores (CS) determined confident classification (CS ≥ 0.9), association (CS 0.3–0.89), or no match (CS < 0.3). Classifier output was correlated with genomic alterations — including MYCN amplification, TERT rearrangements, ATRX mutations, ALT markers (c-circles), and chromosomal copy number alterations — as well as patient survival data.

At the superfamily level, 90% of all 303 samples classified as neuroblastoma with high confidence. At the subclass level, approximately 66% achieved confident classification: MYCN-type, ALT/TERT TMM positive, or TMM negative. Genomic validation in the local cohort confirmed that all TMM-positive classified cases with available data harbored TERT or ATRX alterations or c-circle positivity. Notably, some TERT-rearranged tumors classified as MYCN-type rather than TMM positive, suggesting MNA-driven methylation patterns can override or overlap TERT-associated epigenetic signatures. A subset of MYCN-type cases lacked evident genomic MYCN amplification, potentially implicating activating ALK mutations as drivers of a similar methylation landscape.

Survival analysis showed that MYCN-type and ALT/TERT TMM-positive patients had comparably poor outcomes, both significantly worse than the TMM-negative group. The methylation arrays also enabled inference of chromosomal copy number alterations: 1p deletion and 17q gain were enriched in MYCN-type tumors, while combinations of 11q loss, 3p loss, 7q gain, and 17q gain characterized TMM-positive cases. This raises the possibility that methylation arrays could replace or supplement SNP arrays for standard-of-care genomic profiling.

While the classifier demonstrated strong overall performance, approximately 34% of samples did not achieve confident subclass assignment, and discrepancies between genomic features and methylation classification were observed in a meaningful minority of cases. The authors acknowledge limitations including cohort composition differences, varying array platforms, and the absence of complete TMM data for all samples. Nonetheless, the study establishes DNA methylation-based classification as a promising molecular diagnostic tool for neuroblastoma, with potential to integrate TMM status, CNAs, and risk stratification into a single assay.