Introduction: Increasing evidence demonstrates that the activation states and diverse spectrum of macrophage subtypes display dynamic heterogeneity in the tumor microenvironment, which plays a critical role in a variety of cancer types.
Objectives: To investigate the heterogeneity and the homeostasis of different macrophage subtypes, as well as their effect on biological and clinical manifestations of ovarian cancer (OV).
Method: Integrated immunogenomic analysis of single-cell and bulk tissuetranscriptome profiling was performed to systematically investigate the association between macrophage activation and prognostic and therapeutic efficacy. Consensus clustering analysis was used to define novel macrophage subtypes. An artificial neural network was used to simulate the dynamic activation of macrophages.
Results: The pan-cohort results suggested that high relative infiltration abundance of M0 and M1 macrophages was associated with improved outcome and therapeutic efficacy. However, it was the opposite for M2 macrophages. Unsupervised consensus clustering analysis revealed two OV subgroups characterized by a balance between M0, M1 and M2 macrophages with distinct clinical and immunological behaviors. Finally, a macrophage polarization-derived artificial neural network model was proposed to serve as a robust prognostic factor and predictive biomarker for therapeutic efficacy, which was validated in different independent patient cohorts.
Conclusion: The present study provides a new understanding of macrophage heterogeneity and its association with OV prognosis and underlines the future clinical potential of a macrophage activation model for tumor prevention and treatment.
Keywords: Artificial neural network; Immune microenvironment; Macrophages; Single-cell.
Copyright © 2022. Production and hosting by Elsevier B.V.