RUX-AZA Response in AP/BP-MPN Is Driven By Release of the Differentiation Block in Mutant HSPCs: A Single-Cell Multi-Omic Analysis of Responders on the Phase Ib Phazar Study

Background

The prognosis associated with a diagnosis of accelerated-phase (AP, 10-19% blasts) or blast-phase myeloproliferative neoplasms (BP-MPN, ≥20% blasts) remains dismal, with an unmet need for new therapy options. The phase Ib single-arm PHAZAR trial was designed to determine maximum tolerated dose (MTD), safety and efficacy of the JAK1/2 inhibitor ruxolitinib (RUX) in combination with azacitidine (AZA) in patients (pts) with AP/BP-MPN. This academic trial included serial banking of patient samples before and after treatment, providing a unique opportunity to interrogate the cellular and molecular basis of treatment response in AP/MP-MPN. Here we present the final analysis, including serial genetic and single cell transcriptomic profiling of paired samples from responders and non-responders.

Method

Cohorts of 3-5 patients were enrolled at a fixed AZA dose of 75 mg/m² s/c for 7 days of a 28-day cycle with continuous oral RUX dosed at 10, 15, 20 or 25 mg BD using a Continuous Reassessment Method design. RUX/AZA-ineligible pts were recruited to an observational (obs) cohort. Baseline (BL) and serial (every 3 cycles) bone marrow (BM) samples were analysed by flow cytometry for hematopoietic stem and progenitor cell (HSPC) and myeloid blast epitopes, a myeloid gene panel, and single-cell CITE (Cellular indexing of transcriptomes and epitopes)-seq to define cell states in response to treatment.

Results

58 AP/BP-MPN pts were recruited (n=34 RUX-AZA, n=24 obs). For the RUX-AZA cohort, median age was 72 yrs (range 55-85), 20/34 (59%) were male, and 15/34 (44%) in BP-MPN. Median number of RUX /AZA cycles were 4 (range 1 - 47). MTD was established as 25mg RUX BD as previously reported. 31 pts completed cycle 1 & were response evaluable.

5/31 (16%) pts achieved a complete response (CR) and 5/31 (16%) achieved a partial response (PR) after cycles 3 or 6, with an overall best response rate (CR or PR) of 10/31 (32%). Median response duration across all treatment cycles was 7.2 months (95% CI 2.8 – not reached). Median overall survival was 9.3 months (95% CI 5.7-26.3). For AP-(n=18) and BP-MPN (n=13) pts, 42% (95% CI 18-65%) and 26% (95% CI 6-51%) were alive and leukemia-free at 12 months.

BL immunophenotyping confirmed aberrant HSPC profiles with an expanded CD34+Lin- population. AP/BP-MPN blasts were universally CD34+, CD117+ & expressed HLA-DR (90%) & monocytic markers CD13 (80%) and CD33 (50%), while negative for CD235ab. Documented clinical responses were confirmed by reduction in blast % by flow cytometry.

Serial genotyping established that in non-responders and 80% of responders (including those with normal counts for > 12 months), there was no change in clone distribution nor emergence of new mutations on treatment. We reasoned that clinical response is therefore driven by altered cellular/molecular properties of cells within the mutant clone. In order to explore this further, serial samples at BL, response & relapse from responders with long-term survival (n=6: 5 CR, 1 PR), non-responders (n=6) and healthy controls (HC, n=5) were selected for CITE-seq (n=32 samples, n=123019 cells post QC) to investigate the impact of RUX-AZA on cell type composition and molecular state. Leukemic blasts were highly heterogeneous, arrested at different differentiation stages (monocytic 4/11, myeloid progenitor (prog) in 4/11, megakaryocyte/erythroid prog in 3/11 at BL). All BL samples showed a proportionate reduction in erythroid prog, while the eosinophil, basophil, mast prog population was expanded.

Strikingly, despite no change in clonal burden, responders showed RUX-AZA induced release of the mutant HSPC-associated differentiation block with a notable increase in numbers of T-cells. HSC/MPP populations at the response timepoint were strongly enriched for inflammatory pathways with upregulation of IFN-a, IFN-g, TNFa and TGFb signatures. Blast populations in non-responders at treatment failure and also in responders at relapse showed upregulation of MYC and oxidative phosphorylation pathways, in keeping with increased proliferation and therapy resistance.

Discussion

Response to RUX-AZA in AP/BP-MPN is mediated by increased ability of mutant HSPC to differentiate to mature cell types and not by molecular (clonal) response. Response correlates with altered inflammation-associated gene expression and increased numbers of BM T-cells, providing insights into the possible mechanism of response to RUX-AZA.