Abstract BACKGROUND By integrating whole-genome sequencing (WGS) and RNA-seq data, we found that intracranial germ cell tumors (IGCTs) can be divided into three distinct molecular subtypes: immune-hot, MYC/E2F, and SHH, each characterized by a unique profile of genomic abnormalities. To further define these subtypes, we subsequently performed whole-genome bisulfate sequencing (WGBS) on IGCT tumors METHODS Within our prior discovery cohort (n=110), DNA from 84 samples was available for profiling methylation levels using Whole Genome Bisulfite Sequencing (WGBS). Fastq data was aligned to the human reference genome (hg19), with duplicates removed, and DNA methylation was extracted using bismark pipeline (ver0.24.0). After quality control, weight methylation level was calculated for each CpG island as the sum of methylated reads divided by coverage. To identify IGCTs molecular groups, we applied hierarchical consensus k-means clustering to the most variant methylated CpG island (n=1000) across 84 IGCTs tumor tissues. The methylation level of RNA-seq subtypes was visualized to further clarify the molecular subtypes of IGCTs RESULTS Unsupervised clustering revealed that IGCTs can be divided into three groups based on the CpG island methylation levels: hypomethylation group (n=32), moderate methylation group (n=27), and hyper methylation group (n=25). The hypomethylation and moderate methylation group are predominantly observed in germinoma (>90%). In contrast, hypermethylation group comprises the majority of non-germinomatous germ cell tumors. Furthermore, these subgroups derived from methylation analysis are highly consistent with the molecular subtypes previously defined through gene transcription profile analysis. We observed that the majority of MYC/E2F subtype (24/33) belonged to the hypomethylation group, whereas the majority of immune-hot subtype (14/23) fell into the moderate methylation group. The SHH subtype was the most common within the hypermethylation group (18/26) CONCLUSIONS The three molecular IGCT subtypes showed unique methylation profile. These results indicate different mechanisms of tumorigenesis among the subtypes and pave the way for the clinical application of methylation-based subtyping.
Read full abstract