Integrative Machine Learning of Glioma and Coronary Artery Disease Reveals Key Tumour Immunological Links

Abstract

It is critical to appreciate the role of the tumour-associated microenvironment (TME) in developing strategies for the effective therapy of cancer, as it is an important factor that determines the evolution and treatment response of tumours. This work combines machine learning and single-cell RNA sequencing (scRNA-seq) to explore the glioma tumour microenvironment's TME. With the help of genome-wide association studies (GWAS) and Mendelian randomization (MR), we found genetic variants associated with TME elements that affect cancer and cardiovascular disease outcomes. Using machine learning techniques high dimensional data was analysed to obtain new molecular sub-types and biomarkers that are important for prognosis and treatment response. F3 was identified as a top regulator and revealed potential angiogenic and immunogenic characteristics within the TME that could be harnessed in immunotherapy. These results demonstrate the potential of machine-learning approaches in identifying and dissecting TME heterogeneity and informing treatment in precision oncology. This work proposes improving the immunotherapeutic response through targeted modulation of relevant cellular and molecular interactions.