Background: Pts with HR-MDS after HMA failure have a poor overall survival (OS) of < 6 months. Immune escape is associated with resistance to HMAs in MDS. We hypothesized that CTLA-4 blockade in these pts would be tolerable and lead to clinical responses. Methods: This investigator-initiated, CTEP-sponsored, multi-center phase 1b study enrolled pts after failure of HMAs. In dose-escalation, ipi monotherapy was given at 2 dose levels (DL): 3 and 10mg/kg. Four doses (every 3 weeks) were administered followed by a maintenance phase (4 doses every 3 months) for non-porgressors. Toxicities and responses were evaluated with CTCAE4 and IWG2006 criteria, respectively. OS was estimated using the Kaplan-Meier method. The impact on T-cells were studied by flow cytometry and TCR sequencing. Results: 29 pts from 7 centers were enrolled. Mean age (SD) was 67 (8) years. Most had IPSS high/int-2 (55%), 45% had int-1. Three of 6 pts in DL1 and 4 of 5 pts in DL2 experienced grade [G]2-4 immune-related adverse events [IRAEs] that were reversible with drug discontinuation or systemic steroids. The DL1 (3mg/kg) was expanded with no G2-4 IRAEs reported in the 18 additional pts. A total of 15 deaths occurred due to disease progression or other complications but none attributed to ipilimumab. In total, 52% received all 4 induction doses, and 24% received ≥1 maintenance dose. Best objective responses were 2 marrow complete responses (mCR, 7%). Prolonged stable disease (PSD) for ≥46 weeks occurred in 6 pts (21%) and for ≥54 weeks in 3 pts (10%). Five pts (17%) subsequently underwent allogeneic transplantation (alloSCT) without evidence of excessive toxicity. Median OS (censoring at alloSCT) was 294 days (95%CI, 240-671+) and 400 days (95%CI, 240-671+) for those who received maintenance (n = 7). Pts who achieved PSD and mCR had significantly increased expression of ICOS, a marker of T-cell activation. Conclusions: Immune checkpoint blockade with ipi is tolerable and can lead to PSD/mCR in a proportion of pts. However, ipi monotherapy efficacy is limited after HMA failure and combination-based approaches should be considered. Increased frequencies of ICOS expression might predict clinical benefit. Clinical trial information: NCT01757639