S1615 SAFETY, EFFICACY, AND BIOMARKERS OF RESPONSE TO AZACITIDINE (AZA) WITH NIVOLUMAB (NIVO) AND IPILIMUMAB (IPI), AND AZA WITH NIVO IN R/R ACUTE MYELOID LEUKEMIA: A NON‐RANDOMIZED, PHASE 2 STUDY.

Please indicate where the abstract has been published before: part of this clinical trial (AZA+Nivo) was published in Cancer Discovery 2018. Background: PD‐1 positive CD8 T‐cells were increased in the bone marrow (BM) of patients (pts) with relapsed AML (Williams et al., Cancer 2018). Azacitidine up‐regulates PD‐1 and interferon‐gamma signaling (Yang et al., Leukemia 2013). Aims: To determine safety, response rate and overall survival (OS) for Aza+Nivo and Aza+Nivo+Ipi and identify immunological and molecular biomarkers of response. Methods: Pts were eligible for the AZA+Nivo (cohort 1) if they had relapsed/refractory AML (R/R AML), ECOG ≤ 2, and adequate organ function. 70 R/R AML pts were treated. This cohort is closed. A subsequent cohort of AZA+Nivo+Ipi was opened (cohort 2), with the same eligibility criteria. Ipi 1 mg/kg Q6 weeks was added to the established AZA+Nivo schedule. This dosing was based on lung and melanoma dosing for Ipi + Nivo. Results: Cohort 1 of 70 R/R AML pts treated with Aza+Nivo (Daver et al., Cancer Discovery Nov 2018). Response rates and OS were superior to historic HMA‐based clinical trial controls at MDACC, with most significant OS improvement in patients with low pretherapy BM blast burden (<20% BM blasts) and early salvage (Salvage 1). Similar features predict for response to blinatumomab (Topp M et al., Lancet Oncology 2015) and CART (Park J et al, NEJM 2018) in relapsed B‐ALL, suggesting progressive T‐cell exhaustion with multiple relapses. Responders (Rs) to AZA+Nivo had a higher frequency of pretherapy BMA CD3+ cells compared with non‐responders (NRs) (optimal CD3+ cutoff 13.2%). To further interrogate the functionality of T‐cells in Rs versus NRs we applied a 32‐plex, single cell, stimulated T‐cell, cytokine response panel to pretherapy sorted BM T‐cells. The pretherapy BM T‐cell polyfunctional strength index (PSI) defined as the percentage of polyfunctional cells in the sample, multiplied by the intensities of the secreted cytokines, was dramatically different between Rs and NRs, especially for CD4 cells (p = 0.0317) (Figure 1A), predominantly driven by IFN‐g. All CR/CRi pts, and none of the NR pts, had pretherapy PSI>10 with a statistically significant OS difference for PSI 10 (p = 0.0018) (Figure1B). Cohort 2 of 24 R/R AML pts treated with Aza+Nivo+Ipi with median (med) age 71 years (26–86), secondary AML (46%), poor risk cytogenetics (63%), TP53 (38%), med salvage 2 (range, 1–4) with 54% treated with prior HMA based therapies. 2 pts had prior allo‐SCT. The CR/CRi was 9 (38%), additionally 8 pts (33%) achieved stable disease (SD) (defined as absence of CR, CRi, PR, MLFS; with stable disease on treatment for at least 3 months), and 7 (29%) were NRs. The 4‐week and 8‐week mortality were 0 and 8%, respectively. In all salvage setting the med OS in Aza+Ipi+Nivo versus Aza+Nivo versus historical HMA‐based clinical trial cohort at MDACC, were 10.5, 6.4, and 4.6 months, respectively (P = 0.0025) (Figure 1C). The 1‐year OS in R/R AML with AZA+Nivo+Ipi was very encouraging at 45%. Grade 3/4 immune mediated toxicities were observed in 6 pts (25%), including rash, pneumonitis, and colitis. One pt required ICU stay but no deaths were attributed to immune toxicity. Other grade >2 toxicities were as expected for R/R AML population and were mostly infections/febrile neutropenia, and electrolyte disturbance. Summary/Conclusion: The CR/CRi rates and OS with Aza+Nivo+Ipi are encouraging and study is enrolling. PSI on pretherapy BM CD4 T‐cells from pts treated with Aza+Nivo clearly segregated Rs vs NRs (p = 0.0317) suggesting a very significant and hitherto underappreciated immune diversity in AML and a need for biomarker based trials with immunotherapies in AML.