Abstract Background Brain tumors in young children, represent a significant clinical challenge due to their complex biology, aggressive behavior, and diverse histopathological presentations. Hemispheric high-grade gliomas in infants often exhibit indistinct histological features, making accurate diagnosis difficult using traditional methods alone. Recently, infant-type hemispheric gliomas (IHG) were described as a unique entity characterized by tyrosine kinase alterations including ALK, NTRK1/2/3, ROS1 and MET. Herein, we describe a novel methylation tumor subgroup with overlapping characteristics with infant-type hemispheric gliomas, which show embryonal characteristics. Methods We used dimensionality reduction of a dataset of 27,414 methylation profiles of CNS tumors to identify methylation groups. A small methylation group was identified near the IHG, subgroup, but scored poorly for all established entities, on the Heidelberg and NCI/Bethesda classifiers and clustered robustly as a distinct methylation subclass. Results A total of eighteen samples are included in this cluster, visualization of the genome-wide methylation pattern using uniform manifold approximation and projection (UMAP) confirmed a common pattern with proximity to the IHG methylation group, supporting the unique nature of this molecular subgroup. Median age at diagnosis was 1.4 years (range: 0.1-21), with only one patient older than 4 years of age; nine patients were female. Histologically, this group was enriched by embryonal histological appearance and the presence of ALK and NTRK alterations. Conclusion Using DNA methylation classification, we identified a novel tumor subgroup termed “IHG-EMB”. This group is characterized by supratentorial tumors presenting in young children, frequently less than 4 years of age, with embryonal histology and frequent ALK and NTRK alterations. The distinct and robust clustering of this group suggests a rare new subtype, that may be a variant of the main IHG group of tumors, sharing the features of gene fusions as oncogenic drivers. Cohort expansion to test for robust clinical correlations is ongoing.
Fonseca et al. (2025) studied this question.