A clinically annotated transcriptomic atlas of nervous system tumors

Area of Science:

• Neuroscience
• Genomics
• Oncology

Background:

• Distinct DNA methylation signatures exist across nervous system neoplasms.
• Transcriptomic signature uniqueness and large-scale comparative datasets were lacking.
• This study addresses these knowledge and resource gaps.

Purpose of the Study:

• To determine if transcriptomic signatures are distinct across nervous system neoplasms.
• To create a large-scale dataset for comparative gene expression analysis.
• To refine transcriptional subtypes of specific tumors.

Main Methods:

• Compiled and harmonized transcriptomic and clinical data from public sources (5,402 neoplastic, 1,973 nonneoplastic samples).
• Adjusted for batch effects using surrogate variable analysis.
• Applied machine learning for visualization, clustering, and reclassification of samples with uncertain diagnoses.

Main Results:

• Generated the largest clinically annotated transcriptomic atlas of nervous system tumors.
• Sample clustering was primarily driven by diagnosis.
• Refined pheochromocytoma and paraganglioma (PH/PG) transcriptional subtypes into 6 robust subtypes, validated with TCGA data and correlated with mutational signatures and clinical behaviors.

Conclusions:

• Bulk transcriptomic signatures are distinct across the nervous system neoplasm diagnostic spectrum.
• The atlas provides broad diagnostic coverage, age range, and geographical diversity for comparative analysis.
• The atlas utility is exemplified by PH/PG subtype refinement and analysis.