Applying Statistical Learning Algorithms on the Prediction of Response to Immune Checkpoint Blockade Therapy

• Activity Number: 81 - Contributed Poster Presentations: Section on Statistics in Epidemiology
• Type: Contributed
• Date/Time: Monday, August 3, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Tiantian Zeng* and Chi Wang
• Companies: University of Kentucky and University of Kentucky
• Keywords: machine learning; prediction; immune therapy; statistical modeling

Abstract

Immune checkpoint blockade (ICB) therapy could bring long-lasting clinical gains for the treatment of cancer. However, studies show that only a fraction of patients respond to the treatment. In this regard, the statistical modeling, which constructs classification algorithms to predict patients’ response to the ICB therapy, could help explore the complexity of immune response. In this study, we used several published melanoma datasets with RNA-seq data and clinical response, and built prediction models using random forest and Lasso methods. We found that the specific pairwise relations of the expression of immune checkpoint genes performed the best in predicting the treatment response. In addition, we compared the prediction performance using combined datasets versus each single dataset. Our finding demonstrated that the utilization of statistical modeling and data integration is of high value to identify ICB response biomarkers in future studies.