Abstract Medulloblastoma (MB) is the most common embryonal brain tumor and displays marked heterogeneity, with Group 3/4-MB being the most diverse subgroups. DNA methylation profiling of bulk tumor tissues has been instrumental in understanding the regulatory landscape of MB and enabled precise subtype stratification. However, intra-tumoral epigenetic heterogeneity has remained largely unexplored, prompting us to generate the first single-cell DNA methylation dataset across all MB subgroups and subtypes (31 samples, 5731 cells). Our high-resolution dataset (∼10 % genomic coverage/cell) enabled both epigenetic and genetic characterization. Using neural network classifiers, we assigned methylation subgroup and subtype identities for each cell. Strikingly, the Group 3/4 continuum is evident within individual samples, and we observe multiple subtypes co-existing. Further, we find that subclonal architecture is more prevalent than previously appreciated, with a majority of tumors containing multiple genetic clones. Reconstructing phylogenetic trees allowed us to trace tumor evolution and link progression to epigenetic remodeling. Broadly, we observe a decrease in global methylation levels during tumor progression alongside changes in megabase-scale partially methylated domains. These are further associated with significant changes in gene expression. The combined study of genetic and epigenetic tumor evolution identified the growth hormone PTHLH as a distinctive marker of subtype VIII, the most common of the Group 3/4 subtypes. Subtype VIII-specific hypomethylation patterns of PTHLH correlate with elevated gene expression, likely mediated by STAT5 signaling. In one illustrative case, loss of PTHLH expression accompanies the emergence of a PRDM6-activated, more aggressive tumor state, which is associated with shorter overall survival compared to tumors with high PTHLH expression in an independent patient cohort. Taken together, our study establishes a first-of-its-kind single-cell DNA methylation atlas of MB, enabling dissection of the interplay between genetic and epigenetic changes during tumor evolution and providing novel insights to guide future precision therapies.
Larsson et al. (Tue,) studied this question.