A Phase II Trial of Azacitidine with Ipilimumab, Nivolumab, or Ipilimumab and Nivolumab with or without Azacitidine in Relapsed or Refractory Myelodysplastic Syndrome

Introduction Patients with myelodysplastic syndrome (MDS) who progress on hypomethylating agents (HMAs) have poor outcomes; effective therapies remain an unmet clinical need. Programmed cell death ligand 1 (PD-L1) and cytotoxic T-lymphocyte associated protein 4 (CTLA4) are upregulated in MDS cells after HMA exposure - providing a rationale for assessing the efficacy and safety of PD-L1 and CTLA4 blockade in HMA failure MDS. Here, we present the extended follow-up of three immunotherapy-based approaches. Methods This phase II trial enrolled patients ≥18 years with MDS following progression on a HMA (NCT02530463). Patients received one of three regimens: ipilimumab (ipi), nivolumab (nivo), or ipilimumab-nivolumab (ipi-nivo). Ipi was administered at 3 mg/kg IV D1 in 21-day cycles. Nivo was administered at 3 mg/kg IV on D1 and D15 in 28-day cycles. In the ipi-nivo cohort, ipi was given at 1 mg/kg IV on D1 with nivo at 3 mg/kg IV on D1 in 28-day cycles. For patients with disease progression on immunotherapy, azacitidine (aza) 75 mg/m2 IV was given on D1-5 in 28-day cycles with ipi 3 mg/kg IV on D6 or nivo 3 mg/kg IV on D6. For patients progressing on ipi-nivo, aza was administered in combination with ipi 1 mg/kg IV and nivo 3 mg/kg IV, both on D6 in 28-day cycles. The primary outcome was to determine the overall response (ORR), defined as achievement of complete response (CR), CR with limited count recovery (CRL), or hematological improvement (HI) by IWG 2023. Secondary outcomes aimed to evaluate the duration of response, overall survival (OS), and toxicities. Results We analyzed 55 patients in three cohorts: 20 were treated with ipi, 15 with nivo, and 20 with ipi-nivo. The median age was 70 y. (range: 41-84). Fifteen (27%) had therapy-related MDS, and 28% had complex cytogenetics. The median IPSS-R score was 4 (1-9.5), and the median IPSS-M score was 0.10 (-1.79-4.13). The most common mutations were ASXL1 (35%), TET2 (28%), TP53(24%), and RUNX1 (16%). Patients received a median of 3 cycles of therapy in each cohort. In the ipi cohort, 15% achieved HI and 5% achieved CRuni; the ORR was 20%. In the nivo cohort, no patient responded, and the ORR was 0%. In the ipi-nivo cohort, 11% achieved CRuni and 5% achieved HI; the ORR was 16%. There were no differences in the ORRs between cohorts. The median duration of response for ipi was 4.3 m. and 7.9 m. for ipi-nivo (p = 0.111). In a pooled cohort analysis, DNMT3A and ASXL1 mutations were associated with response achievement (DNMT3A OR: 11.82, p = 0.027; ASXL1 OR: 6.98, p = 0.014). TET2 mutations were associated with lack of response (OR: 0.12, p = 0.036). In patients that progressed on immunotherapy, the best response for all patients treated with azacitidine augmentation was marrow CR (mCR). The proportion of patients achieving mCR was 30% for aza-ipi, 17% for aza-nivo, and 33% for aza-ipi-nivo. After pooling the cohorts and accounting for mCRs, azacitidine augmentation failed to improve responses (ORR + mCR) compared to immunotherapy alone (26% vs 27%, p > 0.999). The median OS was 8.9 m. for ipi, 8.0 m. for nivo, and 8.7 m. for ipi-nivo (p = 0.620). In a pooled multivariate Cox model, complex karyotypes, normal karyotypes, and RUNX1 were associated with significantly improved OS (MDACC favorable risk); TP53, TET2, monosomy 5, and del(7q) were associated with significantly worse OS (MDACC adverse risk). We stratified the hazard ratios to create an immunotherapy-specific prognostic classification for HMA failure MDS. The OS of the MDACC favorable risk category was 21.2 m., superior to the MDACC adverse risk category at 7.1 m. (p = 0.013). Grade ≥3 neutropenia and lymphocytopenia were more frequent with ipi-nivo compared to ipi or nivo (neutropenia: 90% vs 53%, p= 0.022; lymphocytopenia: 80% vs 43%, p = 0.011). Ipi-nivo was associated with longer hospitalization compared to ipi or nivo combined (6 d. vs. 1 d., p = 0.01). The incidence of immunotherapy-related adverse events (irAEs) across treatment arms was 44%, occurring more frequently with ipi-nivo. Dermatologic or endocrine irAEs were associated with improved OS compared to patients with no irAEs (13.9 m. vs 7.0 m., p = 0.030). Conclusion PD-L1 and CTLA4 blockade are safe in HMA failure MDS, but are associated with only modest clinical activity. Immunotherapy failed to resensitize patients to HMAs. Dermatologic and endocrine irAEs are associated with improved OS. Unique cytogenetic and molecular signatures may be biomarkers for patient selection.