Explainable machine learning to identify chronic lymphocytic leukemia and medication use based on gut microbiome...

Area of Science:

• Microbiome Research
• Host-Microbe Interactions
• Clinical Metagenomics

Background:

• Medications, especially antibiotics, significantly impact gut microbiome diversity and composition.
• The gut microbiome is increasingly recognized for its role in influencing cancer progression and overall host health.
• Understanding microbiome variations in disease states is crucial for developing targeted therapeutic strategies.

Purpose of the Study:

• To investigate gut microbiome composition across various diseased and healthy cohorts.
• To assess the impact of medication, disease, age, and sex on microbiome variation.
• To validate and refine machine learning-based microbiome signatures associated with chronic lymphocytic leukemia (CLL).

Main Methods:

• Integration of clinical, shotgun metagenomic, and medication data from diverse patient cohorts and healthy individuals.
• Comparative analysis of microbiome diversity and composition across cohorts including CLL, AML, MDS, cardiac surgery patients, and kidney donors.
• Application of machine learning algorithms to identify and validate disease-specific microbiome patterns.

Main Results:

• Similarities in gut microbiome diversity and composition were observed between patients with chronic lymphocytic leukemia (CLL) and those scheduled for elective cardiac surgery.
• Patients with acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS) exhibited the least diverse and most distinct microbiomes.
• Medication, disease, age, and sex collectively explained only a small fraction (4%-10.4%) of microbiome variation, with 90%-95% remaining unexplained.

Conclusions:

• Disease status alone has a limited impact on shaping microbiome composition, as evidenced by similarities between CLL and cardiac surgery patient cohorts.
• The significant unexplained microbiome variation underscores the need for identifying novel factors influencing gut microbial communities.
• Validation of the CLL-associated microbiome signature confirms its robustness and potential for clinical application in managing CLL.