Abstract Background The programmed death ligand 1 (PD-L1) has recently emerged as a target immunotherapy in Triple-negative breast cancer (TNBC). However, the tumor-intrinsic role of PD-L1 and its pathway still needs to be fully clarified. Recently, a close association between CD73, PD-L1, cancer cell steaminess and epithelial-to-mesenchymal transition phenotypes is emerging. In addition, we have previously demonstrated that miR-320a and miR-145 target PD-L1, whereas miR-30 family is known to target CD73. In this study, we aimed at investigating the role of these miRNAs as prognostic and predictive dynamic biomarkers for immunotherapy in TNBC. Additionally, we investigated whether PD-L1 exerts cellular autonomic functions in TNBC beyond its role in the immune checkpoint. Results CD73 and PD-L1 expression was assessed in a panel of breast cancer lines and the non-malignant breast cells MCF-10. We showed a significantly higher CD73 and PD-L1 expression in MDA-MB-231 (TNBC, Basal B) than in MCF-10. Conversely, in MCF-7 cells (HR +, Luminal A), CD73 and PD-L1 expression is lower than in MCF-10. Furthermore, we showed that low PDL1 expression is correlated with high levels of miR-320a and miR-145, and CD73 expression is inversely related to miR-30 expression. These putative biomarkers were also validated in two patients with different PD-L1 status. From the analysis of the data obtained by real-time qRT-PCR, we found high plasma level of CD73 and low plasma level of miR-320a, miR-145 and miR-30 in the patient with PD-L1 positive BC. On contrary, in the patient with PD-L1-negative BC, we found low plasma level of CD73 and high plasma level of miR-320a, miR-145 and miR-30. Furthermore, to explore whether PD-L1 could also have an intrinsic role in tumor development and invasiveness, we used stably MDA–MB-231 PD-L1-silenced cells. We found that cell growth, colony formation, migration rate and the ability to form spheres are consistently reduced upon shRNA-mediated PDL1 silencing. In particular, we observed a reduction in the number, diameter and volume of the 3D spheres compared to the control cells. To further demonstrate the role of PD-L1, we treated the PDL1-high expression cells MDA-MB-231, PD-L1 silenced MDA-MB-231 clones, and PD-L1 low expression cells MCF-7 with Durvalumab, an anti-PD-L1. Interestingly, we observed that in PD-L1 silenced clones and MCF-7 cells, the sphere-forming ability was increased in presence of Durvalumab treatment. On the contrary, in MDA-MB-231, Durvalumab inhibited the sphere-forming ability. Conclusion Our data confirm that PD-L1 and CD73 are targets of miR-320a/miR-145 and miR-30 respectively. These could be potential predictive dynamic biomarkers for chemotherapy, ICIs and anti-CD73 therapeutic approach. Future validations of these biomarkers in an extensive series of TNBC patients are needed to support their use in clinical practice. Here, we further characterized the cellular autonomic role of PD-L1 in breast cancer and showed a differential role of basal PD-L1 expression in PD-L1 checkpoint inhibitors treatment efficacy. This suggests a potential role in monitoring PD-L1 expression indirect biomarkers (i.e. miR-320a, miR-145 and CD73) during ICIs treatment.