Skip to main content

DNA Methylation Data Helps Refine Classification of Central Nervous System Tumors

NEW YORK (GenomeWeb) – An international team led by German investigators has come up with a DNA methylation-based classification scheme for classifying central nervous system (CNS) tumors.

Researchers from the National Center for Tumour Diseases (NCT) in Heidelberg, the German Cancer Research Center (DKFZ), the University Hospital Heidelberg, and elsewhere did array-based methylation profiling on tumor and matched normal tissue samples from more than 2,800 individuals with cancer. The set spanned roughly 100 CNS tumor types, along with a range of solid tumors and/or blood cancer types.

From these data, the team defined 82 methylation-based CNS tumor classes, encompassing new and known tumor groups. The results, appearing online today in Nature, suggested that methylation profiling may offer an avenue for expanding and improving CNS tumor diagnoses — a notion supported by diagnostic classifications done on more than 1,100 prospectively collected CNS tumor cases.

The group's methylation-informed CNS tumor classification system, which they heralded as a "platform for next generation neuropathology," is freely available online.

"A uniform implementation of the classification algorithm holds great promise for standardization of tumor diagnostics across centers and across clinical trials," DKFZ researchers Stefan Pfister and Andreas von Deimling, the study's co-corresponding authors, and their colleagues wrote. "Furthermore, the digital nature of methylation data facilitates easy exchange and will allow aggregation of extensive tumor libraries."

The researchers began by using Illumina Infinium HumanMethylation 450 BeadChip arrays to profile genome-wide methylation patterns in freshly frozen or formalin-fixed, paraffin-embedded tumor samples from 2,801 individuals with CNS tumors, mesenchymal tumors, melanoma, diffuse large B-cell lymphoma, plasmacytoma, or pituitary adenomas.

Through an iterative clustering analysis, the team identified 82 methylation-informed CNS clusters. More than two-thirds of these clusters overlapped with existing tumor class or sub-class designations from the World Health Organization, while the remaining tumors differed to varying degrees from WHO classification schemes.

The researchers subsequently incorporated CNS tumor data from the Cancer Genome Atlas and continued to refine their methylation classifier to make it more adept at doing the sorts of rapid and reproducible tumor classification that's needed in clinical diagnostics.

When prospectively applying it to 1,155 more CNS tumors, the team got informative methylation profiles for all but 51 cases, and assigned 977 of the tumors to methylation classes. While the majority of those groups lined up with histology-based classifications for the tumors, the authors noted that the methylation-informed approach "will probably result in the detection of exceptionally rare tumor classes and a continued refinement of classifiers" and "more dynamic tumor classification." 

The use of DNA methylation signatures in combination with histology and molecular tumor classification "will improve diagnostic accuracy not only in neuropathology, but will serve as a blueprint in other fields of tumor pathology," they added.