P720: a bone marrow microenvironmental cell signature scoring system independently predicted survival and drug resistance in chronic myelomonocytic leukaemia patients

Background: Chronic myelomonocytic leukaemia (CMML) is a clonal myeloid malignancy arising from haematopoietic stem cells. The clinical features and outcomes of CMML patients vary considerably, underscoring the importance of identifying genetically and clinically distinct subgroups. Emerging evidence suggests that various cellular components and associated signaling pathways in the bone marrow (BM) microenvironment play critical roles in the pathogenesis of CMML. Aims: This study aims to investigate the clinical relevance of BM cell signatures in CMML patients. Methods: We enrolled 75 CMML patients who were diagnosed and treated at the National Taiwan University Hospital and had adequate cryopreserved BM mononuclear cells (MNCs) for RNAseq analysis, correlated with comprehensive clinical metadata from presentation. Single-sample gene-set enrichment analysis (ssGSEA) was adopted to score signatures of 20 distinct cell types in the BM. Results: Signatures of seven cell types, including granulocyte-monocyte progenitors, macrophages, basophils, neutrophils, platelets, CD4+ T cell, and CD8+ T cell, were significantly associated with overall survival (OS) by backward conditional Cox regression approach. A BM cell signature (BCS-7) scoring system was constructed based on the weighted sums derived from Cox regression analysis. High BCS-7 scores correlated with higher scores in the clinical/molecular prognostic scoring system in CMML (CPSS-Mol, Elena et al, Blood 2016) (p=0.007), higher rates of mutations in tumour suppressor genes (20% vs 2.2%, p=0.009), lower rates of splicing gene mutations (33.3% vs 57.8%, p=0.038), and importantly, clinical non-responsiveness to hypomethylating agents (HMAs) (70% vs 23.1%, p=0.024). High-score patients had significantly inferior leukemia-free survival (LFS) and OS than low-score patients (median, 7 vs 30.5 months, and 11.2 vs 38.4 months, respectively, Figure 1). In subgroup analyses the BCS-7 score retained strong predictive value for LFS and OS in CMML-1, CMML-2, and in dysplastic and proliferative subgroups. Furthermore, the prognostic significance of BCS-7 scores for LFS and OS remained valid across CPSS lower-risk and higher-risk groups ASXL1 mutations have been consistently associated with poor prognosis in CMML and have been incorporated into current prognostic systems. Interestingly, high BCS-7 score remained prognostically detrimental in both ASXL1-unmutated and ASXL1-mutated groups. Time-dependent ROC curve analysis showed increased area under curve when BCS-7 was incorporated into CPSS-Mol (Figure 1) suggesting potential for the BCS-7 score to improve current risk stratification systems. The prognostic implication of the BCS-7 score was further validated in two independent cohorts; RNAseq datasets derived from MNCs of 25 patients, and from sorted HSCs from 42 patients, respectively. Multivariable analysis showed that higher BCS-7 score remained an independent adverse risk factor for LFS and OS, irrespective of age and CPSS-Mol risk. Finally, corresponding to our clinical observation, ssGSEA highlighted remarkably higher HMA resistance signatures in BCS-7-high patients, potentially partially explaining the dismal prognosis in this population. Conclusion We describe a novel BM cell signature (“CMML-BCS-7”) system based on various cell signatures in the BM microenvironment of CMML patients, which could serve as an independent prognostic predictor for this heterogeneous disease. Prospective studies are warranted to confirm its clinical significance, and particularly to decipher the differences of BM microenvironment and its association with drug resistance between patients with high and low scores. Figure 1.Keywords: Prognosis, CMML, Signaling, Bone marrow microenvironment