PDL1/CD274 gain/amplification as a predictive marker of checkpoint blockade inhibitor efficacy in metastatic breast cancer: Exploratory analysis of the SAFIR02-IMMUNO randomized phase II trial

PD(L)1 inhibitor have shown efficacy for limited sub population of patients (pts) with HER2 negative metastatic breast cancer (MBC). The main predictive marker of efficacy to date are the absence of ER and PR receptor, and pdl1 positivity by IHC. We investigated copy number alteration (CNA) of the PDL1 gene (also named CD274) located at 9p24.1 in the SAFIR02 Breast Immuno randomized phase II trial (NCT02299999). SAFIR02 BREAST IMMUNO randomized 199 pts presenting a MBC without actionable genomic alterations, responding to 6 months standard chemotherapy, either on durvalumab (10 Mg/kg every two weeks) or on maintenance chemotherapy with a 2:1 ratio. Eighty-two (43%) pts had a triple negative (TN) MBC. Using metastatic tumor samples, PDL1 CNA were characterized from array CGH analysis (Affymetrix CytoscanHD or Oncoscan). A gain of copy number was defined as 3–4 copies and an amplification ≥ 5 copies. Treatment effect was estimated in each subgroup using a cox proportional hazard model. For PDL1 CNA were available for 153 pts (101 immuno, 52 chemotherapy). PDL1 copy loss, neutral, or copy gain/amplification were reported on 30 (20%), 93 (61%) and 30 (20%) of pts, respectively. Pts with TN MBC had a higher proportion of gain/amplification (23/65 pts, 35% for TN tumors; vs 7/82, 8.5% for non-TN). Improvement of OS with durvalumab was limited to the PDL1 CNA gain/amplification subgroup (HR = 0.17, 95% CI 0.05-0.55) with a median OS of 9 months (95%CI 4-18) in maintenance arm and not reached in durvalumab arm. Among pts with TN tumors, durvalumab was associated to a better OS in the gain/amplification subgroup (HR 0.18, 95%CI 0.05-0.71 ), compared to the neutral/loss subgroup (HR 1.1, 95%CI 0.47-2.6 ). This exploratory subgroups analysis of the first randomized trial comparing a PDL1 inhibitor to chemotherapy in the maintenance setting shows that PDL1 CNA could be an important predictive marker for PD(L)1 inhibitors efficacy. If confirmed on larger series, it could have an important implication on the development of immunotherapy for MBC pts, in particular for subgroups with low immunogenicity such as the luminal subtype.