T-cell immune checkpoint inhibition plus hypomethylation for locally advanced HER2-negative breast cancer: a phase 2 neoadjuvant window trial of decitabine and pembrolizumab followed by standard neoadjuvant chemotherapy

BACKGROUND: Higher levels of tumor-infiltrating lymphocytes (TILs) in breast cancers are associated with increased likelihood of pathologic complete response (pCR) to chemotherapy. DNA methyltransferase inhibitors (DNMTi) can augment immune responses to cancers, decreasing myeloid-derived suppressor cells (MDSCs) and increasing T lymphocyte responsiveness. We have shown that the DNMTi decitabine augments the effectiveness of immunotherapy using murine triple-negative breast cancer (TNBC) models. The primary objective was to determine whether DNMTi+immune checkpoint blockade would increase stromal TIL (sTIL) in primary breast cancers before neoadjuvant chemotherapy (NCT). METHODS: ×4 doses over 5 days) followed by 2 doses of pembrolizumab (200 mg, 2 weeks apart)-before starting NCT. Biopsies before and after window immunotherapy quantified TILs and programmed death-ligand 1 (PD-L1) expression. Patients proceeded to NCT and tumor resection per standard of care. Mid-study, results of the KEYNOTE 522 trial led to patients with TNBC receiving additional pembrolizumab concurrently with standard NCT and in the adjuvant setting. RESULTS: 46 patients (median age 54.5 years, range 28-72; 71.7% white, 28.3% black; 100% female) were treated. 21 patients had TNBC and received neither neoadjuvant pembrolizumab concurrently with NCT nor adjuvant pembrolizumab (Cohort A), 7 patients had TNBC and did receive concurrent and/or adjuvant pembrolizumab (Cohort A2), and 18 patients were estrogen receptor positive and/or progesterone receptor positive and received neither concurrent nor adjuvant pembrolizumab (Cohort B). Blood samples collected after decitabine administration before pembrolizumab showed a 59% decrease (p<0.01) in monocytic MDSCs compared with baseline. 38 patients had paired biopsies for sTIL and 37 for PD-L1 evaluation. Cohorts A/A2 experienced an sTIL increase of 6.1% (p<0.008); Cohort B experienced an sTIL increase of 8.3% (p=0.006). PD-L1 expression increased by 73.9% (p<0.01). 14 of 43 patients (32.6%) who proceeded to resection achieved pCR (n=11 of 27 (40.1%) in Cohorts A/A2 and n=3 of 16 (18.8%) in Cohort B). The most frequently reported immune-related adverse events were adrenal insufficiency (AI) (n=6, 13.0%), maculopapular rash (n=3, 6.5%), and hypothyroidism (n=3, 6.5%). Five of the six AI instances were at least partially attributable to hypophysitis/pituitary dysfunction, and one remains uncertain. CONCLUSIONS: Treatment in the pre-neoadjuvant window with decitabine and pembrolizumab could sensitize breast cancers to standard NCT by recruitment of TILs to the tumor tissue. The treatment was well-tolerated. TRIAL REGISTRATION NUMBER: NCT02957968.