Background: Clinical trials of immune checkpoint inhibitors in epithelial ovarian cancer (EOC) have not shown clear survival benefit, likely due to the complex immunosuppressive mechanisms of the EOC tumor microenvironment. Still, certain patients experience long-term treatment benefit. However, we lack reliable biomarkers for distinguishing dominant immunosuppressive mechanisms and for identifying patients with EOC who are responsive to immunotherapy. The present high-dimensional single-cell study analyzed patients with relapsed EOC enrolled in arm A of the NSGO-OV-UMB1/ENGOT-OV30 trial, wherein the patients underwent combination oleclumab (anti-CD73) and durvalumab (anti-PD-L1) immunotherapy. The objective of the study was the identification of blood-based immunophenotypic signatures conducive to the development of improved strategies for patient selection, response monitoring, and personalized targeting of immunosuppressive mechanisms. Methods: A 40-marker suspension mass cytometry panel was utilized for comprehensive phenotypic and functional characterization of longitudinally sampled peripheral blood leukocytes from patients. Artificial neural network-based unsupervised clustering and manual metacluster curation were used to identify leukocyte subsets for differential discovery and correlation analyses. Results: At baseline, short-term and long-term survivors differed with regard to the relative abundances of total peripheral blood mononuclear cells (PBMCs). We observed a significant increase in CD14+CD16- myeloid cells during treatment, initially driven by classical monocyte proliferation and subsequently driven by the expansion of monocytic myeloid-derived suppressor cells (M-MDSCs). This M-MDSC expansion occurred only in patients with shorter progression-free survival, who also showed a continuous decrease in central memory T-cell abundances after baseline. Throughout treatment, we observed upregulation of PD-L1 expression on most T-cell subsets in all patients. Higher expression of CD73 and IDO1 on select leukocyte subsets at baseline was significantly positively correlated with longer progression-free survival. Conclusions: Our study delineates the phenotypic and functional alterations in peripheral blood leukocytes occurring during combination oleclumab/durvalumab immunotherapy in patients with relapsed EOC. We propose a set of biomarkers with potential for treatment personalization and response monitoring: relative abundances of PBMCs at baseline, relative abundances of M-MDSCs and central memory T cells during treatment; PD-L1 expression levels over time; and baseline expression of CD73 and IDO1 on specific leukocyte subsets. However, validation of these biomarkers through larger-scale studies is required.