In silico purification improves DNA methylation-based classification rates of pediatric low-grade gliomas

Area of Science:

• Oncology
• Genomics
• Bioinformatics

Background:

• DNA methylation-based classification is crucial for central nervous system (CNS) tumor molecular diagnosis.
• Pediatric low-grade glioma (pLGG) samples often lack confident classification due to low tumor cell content.

Purpose of the Study:

• To develop a computational method for purifying DNA methylation profiles.
• To enable classification of previously non-classifiable pLGG samples.

Main Methods:

• A reference-based in silico purification framework was developed.
• Epigenetic signatures of five non-malignant cell types were systematically removed.
• Paired DNA methylation profiles from the same biopsy were analyzed for validation.

Main Results:

• Purification enabled confident classification of previously non-classifiable pLGG samples.
• 100% concordance was achieved between initially classifiable and newly classifiable samples post-purification.
• The method successfully classified 24.1% and 22.7% of previously non-classifiable cases in two independent cohorts.

Conclusions:

• The in silico purification framework facilitates accurate molecular diagnosis of pLGG.
• This approach supports timely clinical decision-making for pLGG patients.
• The method shows potential for broader application in other low-purity tumor types.