Classifying Germinal Center Derived Lymphomas-Navigate a Complex Transcriptional Landscape

Area of Science:

• Hematology
• Molecular Biology
• Bioinformatics

Background:

• Lymphoid neoplasm classification relies on histology, immunology, and genetics, augmented by gene expression profiling (GEP).
• Existing GEP classifiers for lymphoma subtypes face challenges with competing molecular signatures and unclear relationships for rare entities like double-hit lymphomas.
• The 5th edition of the WHO classification (2022) aims to refine diagnostic categories for hematologic tumors.

Purpose of the Study:

• To explore the transcriptional landscape of a wide spectrum of lymphomas using machine learning.
• To identify transcriptional states and their relationship to B-cell progression in the germinal center.
• To develop and evaluate novel molecular classifiers for lymphoma subtyping.

Main Methods:

• Application of self-organizing maps (SOM) machine learning to analyze gene expression data from 873 lymphomas.
• Analysis of transcriptional states and their correlation with germinal center B-cell progression.
• Extraction and comparison of gene signatures from identified transcriptional states against existing literature classifiers.

Main Results:

• A transcriptional continuum was observed across lymphoma subtypes, without sharp distinctions, paralleling active germinal center B-cell states.
• Germinal center trajectory expression patterns effectively discriminated between lymphoma subtypes.
• Rare lymphoma subtypes were positioned intermediately between major classes like Burkitt lymphoma (BL), diffuse large B-cell lymphoma (DLBCL), and follicular lymphoma (FL).
• Derived gene signatures demonstrated competitive performance compared to existing classifiers and offered functional insights.

Conclusions:

• Lymphoma classification may benefit from viewing transcriptional states as a continuum rather than discrete categories.
• Germinal center B-cell progression provides a valuable framework for understanding lymphoma transcriptional heterogeneity.
• The developed classifiers offer a functional approach to lymphoma diagnostics, potentially enabling personalized medicine strategies for germinal center-derived lymphomas.