Background: Immune checkpoint inhibitors have revolutionized cancer therapy, yet have limited efficacy in estrogen receptor (ER)⁺ breast cancer. Implicated factors include scarcity of tumor infiltrating lymphocytes (TILs), low PD-L1 expression, female gender and liver involvement. In vitro and in vivo studies suggest that epigenetic modulation with HDAC inhibitors modulate regulatory T cells (Treg) and change TIL composition which is further associated with the presence of a specific immune signature. Methods: Patients (pts) with (ER)⁺ metastatic breast cancer, who progressed on multiple prior therapies, were treated with tamoxifen in combination with vorinostat and pembrolizumab either immediately or after 3 weeks of epigenetic priming in a phase II trial. Comprehensive flow-cytometric immunophenotyping, PD-L1 staining and histone acetylation were evaluated on tumor and blood cells. Results: 34 patients (median age 56 years (32-81), heavily pretreated with a median of 5 (2-13) prior regimens received at least one dose of vorinostat and evaluable for response. Grade ¾ toxicities in 2 pts each (6%) included immune hepatitis, fatigue and thrombocytopenia. Grade 2 toxicities were pneumonitis and colitis in 3%, fatigue in 27% and thrombocytopenia in 12% pts. Six patients did not receive pembrolizumab due to rapid progression or toxicity. Clinical benefits rate defined as CR, PR and stable disease > 6m was seen in 5/28 (18%) pts. Tumor lymphocyte infiltration and PD-L1 expression were low. High expression of PD-1/CTLA-4 dual staining in CD8 cells of > 20% in tumor or blood was seen in 5 pts overall, 4/5 (80%) patients with benefit and in one other patient, who was withdrawn due to immune hepatitis in week 3. Both tumor and peripheral blood CD8 PD-1/CTLA-4 dual expression strongly correlated with time to progression and reduction in Foxp3⁺/CTLA-4^high Treg population in tumors by vorinostat. Conclusions: Our data highlight the potential for patient selection based on exhausted CD8+ cells in blood or tumor to predict response to immune checkpoint inhibitors in a small subset of (ER)⁺ breast cancer patients. Clinical trial information: NCT02395627