Machine learning-based prediction of immune infiltration patterns in extrahepatic cholangiocarcinoma

Area of Science:

• Oncology
• Immunology
• Bioinformatics

Background:

• Immune cell infiltration significantly influences the tumor microenvironment in cholangiocarcinoma.
• Current methods for classifying immune infiltration patterns from transcriptomic data are limited.
• Understanding immune heterogeneity is crucial for developing effective cancer therapies.

Purpose of the Study:

• To develop a machine learning framework for predicting immune infiltration patterns in extrahepatic cholangiocarcinoma (EHCC).
• To leverage transcriptomic data for classifying distinct immune profiles within EHCC tumors.

Main Methods:

• Transcriptomic data from 182 EHCC samples were analyzed using CIBERSORT to quantify immune cell proportions.
• K-means clustering identified two distinct immune infiltration patterns.
• Differentially expressed genes (DEGs) and functional enrichment analyses (GO, KEGG, GSEA) characterized cluster differences.
• A machine learning framework involving nine classifiers and feature selection methods was built and validated.

Main Results:

• Two distinct immune infiltration patterns were identified in EHCC, showing significant differences in immune cell composition.
• 116 differentially expressed genes were found between the two immune patterns.
• Functional enrichment revealed differences in immune signaling and translation.
• A Gradient Boosting Machine (GBM) feature set with a Support Vector Machine (SVM) classifier achieved high performance (AUC 0.9958, accuracy 96.7%).

Conclusions:

• A novel transcriptome-based machine learning framework effectively classifies EHCC immune infiltration patterns.
• This framework enhances the understanding of immune heterogeneity in EHCC.
• The study provides a computational foundation for future clinical validation of immune profiles in EHCC.