Identifying cancer-associated leukocyte profiles using high-resolution flow cytometry screening and machine learning

Area of Science:

• Immunology
• Oncology
• Bioinformatics

Background:

• Machine learning (ML) aids clinical decision-making by modeling complex data.
• Identifying novel biomarkers can enhance ML effectiveness in cancer detection.
• Cancer-related immune cell signatures are a promising area for biomarker discovery.

Purpose of the Study:

• To investigate blood immune cell signatures as potential biomarkers for cancer progression.
• To develop and test a pipeline for identifying cancer-specific leukocyte biomarkers.

Main Methods:

• Developed a multiparameter cell-surface marker screening pipeline.
• Utilized flow cytometry for high-resolution leukocyte population profiling.
• Employed CATboost machine learning models for cancer prediction.

Main Results:

• Identified a signature of blood leukocyte subsets, including monocyte subsets.
• Demonstrated that these signatures can predict the presence and type of cancer in murine models.
• Successfully trained ML models to correlate immune cell profiles with cancer status.

Conclusions:

• Blood immune cell signatures show potential as robust biomarkers for cancer detection and characterization.
• The developed screening pipeline can be adapted for use with human cancer patient blood samples.
• This approach offers a novel strategy for enhancing ML-based cancer diagnostics.