A Phase II Trial of Azacitidine with Ipilimumab, Nivolumab, or Ipilimumab and Nivolumab in Previously Untreated Myelodysplastic Syndrome

Introduction

Hypomethylating agents (HMAs) and allogeneic stem cell transplant (SCT) remain the standard of care for high-risk myelodysplastic syndrome (MDS). However, half of patients do not respond to HMA monotherapy. As programmed cell death ligand 1 (PD-L1) and cytotoxic T-lymphocyte associated protein 4 (CTLA4) are up-regulated in MDS after HMA failure, we explored checkpoint blockade in combination with azacitidine. Here, we present the mature data of three immunotherapy-based HMA combinations.

Methods

This phase II trial enrolled 66 patients with previously untreated MDS (NCT02530463). Eligibility criteria included patients ≥18 years with treatment-naive MDS; this trial allowed any IPSS-R risk MDS. Patients received one of three combinations: azacitidine with ipilimumab (aza-ipi), nivolumab (aza-nivo), or ipilimumab-nivolumab (aza-ipi-nivo). Azacitidine 75 mg/m² IV was given on D1–5 in 28-day cycles with ipilimumab 3 mg/kg IV on D6 or nivolumab 3 mg/kg IV on D6 and D20. Patients in the triplet cohort of aza-ipi-nivo received azacitidine D1-5 followed by ipilimumab 3 mg/kg IV and nivolumab 3 mg/kg IV on D6.

The primary efficacy outcome was to determine the overall response, defined as achieving complete response (CR), CR with limited count recovery (CR_L), or hematological improvement (HI) by the IWG 2023 criteria. Secondary outcomes aimed to evaluate overall survival, toxicity, and the proportion of patients who underwent SCT.

Results

We analyzed 66 patients treated in three cohorts: 33 with aza-ipi, 20 with aza-nivo, and 13 with aza-ipi-nivo. The median age was 68 years (range: 39 – 86). Fifty (76%) patients were transfusion-dependent before starting therapy. Significantly more patients had IPSS-R very poor-risk MDS than any other risk category (47%; p = 0.042). TP53 was the most common mutation — more frequent than any other mutation (46.2%; p = 0.003). The median IPSS-M score was very high (1.59; range: –1.55–4.08). Patients received a median of 4 (1–12) cycles of aza-ipi, 6 (2–76) cycles of aza-nivo, and 3 (1–7) cycles of aza-ipi-nivo.

The composite complete remission rate (CRc: CR + CR_L) was 13.3% for aza-ipi, 40% for aza-nivo, and 46% for aza-ipi-nivo. The CR rate significantly favored aza-nivo compared to aza-ipi (40% vs 7%, respectively, p = 0.009). The overall response rate (ORR: CR + CR_L + HI) was 27% for aza-ipi, 55% for aza-nivo, and 54% for aza-ipi-nivo. There was no significant difference in the ORRs between cohorts. Multiple logistic regression identified normal karyotypes were associated with improved ORRs (OR 9.39, p = 0.036), while DNMT3A^mut was associated with reduced ORRs (OR 0.09, p = 0.043) in patients with blasts ≥5%. Seventeen (26%) patients underwent SCT, with no significant differences in transplant rates between cohorts.

Median OS was 22.7 m. for aza-ipi, 14.3 m. for aza-nivo, and 11.8 m. for aza-ipi-nivo (p = 0.666). In a landmark analysis, aza-nivo was associated with superior overall survival for patients with IPSS-R intermediate-risk MDS compared to aza-ipi (NR vs 25.6 m., p = 0.049). Patients with IPSS-R intermediate-risk MDS were associated with significantly improved survival compared to high or very high-risk disease (77.7 m. vs 10.9 m., p < 0.001). The triplet cohort of aza-ipi-nivo was associated with shorter post-transplant OS compared to the doublet cohorts (13.7 m. vs 49.6 m., p = 0.008). In a pooled analysis, a multivariate Cox model using SCT as a covariate identified improved survival in patients with normal karyotypes (HR 0.31, p = 0.045) and inferior survival with TP53^mut (HR 22.55, p < 0.001).

The triplet cohort was associated with more leukopenia 100% vs 58%, p = 0.004), neutropenia (100% vs 70%, p = 0.022), lymphocytopenia (85% vs 42%, p = 0.019), and neutropenic fever (46% vs 19%, p = 0.067) than the doublet cohorts. Twenty-nine (44%) patients had an immunotherapy-related adverse event. The incidence of grade ≥2 pneumonitis was associated with significantly reduced OS (7.4 m. with pneumonitis vs. 22.7 m. without, p = 0.025) and enrichment in TP53^mut.

Conclusion

There are unique cytogenetic and molecular predictors of response and survival with immunotherapy-based approaches in previously untreated MDS. Aza-nivo was associated with improved CR and — for IPSS-R intermediate-risk MDS — overall survival compared to aza-ipi. Aza-ipi-nivo failed to improve response or OS and was associated with higher rates of toxicity and post-transplant mortality.