Background Dysregulated innate immune signaling has been identified as a driver of the complex pathophysiology of myelodysplastic syndromes (MDS), predominantly in low risk (LR)-MDS (Basiorka et al., Blood 2016). The pharmacological targeting of inflammation may therefore help to retard disease development. However, since MDS is a heterogeneous disease, a personalized approach is likely to be necessary, requiring a more detailed knowledge of distinct inflammatory patterns and their variation between genetically defined subgroups of LR-MDS patients (Winter et al., J Clin Oncol 2020). Aims The aim of this work was to describe inflammation states in the bone marrow of LR-MDS patients (IPSS-R ≤ 3.5) compared to controls with and without clonal hematopoiesis of indeterminate potential (CHIP) and to high-risk (HR)-MDS patients (IPSS-R ≥ 4). Methods Cryopreserved bone marrow mononuclear cells (BM-MNCs) and corresponding bone marrow plasma samples from non-CHIP controls (n = 15), CHIP individuals (n = 12) and patients with LR-MDS (n = 47) or HR-MDS (n = 14) were obtained from the MDS registry and the BoHemE study (NCT02867085) at the University Hospitals in Dresden and Leipzig. Targeted expression analysis of 11 inflammasome-related genes (S100A9, NLRP3, PYCARD, CASP1, IL1B, IL18, NLRP1, NLRC4, AIM2, CASP3, CASP4) and multiplex cytokine measurements were performed. To investigate heterogeneity in the LR-MDS cohort, a principle component analysis (PCA) was performed and a cumulative gene expression score was calculated as summed log2 fold changes (mean non-CHIP = 0) of all 11 mRNA expression values for each sample. Furthermore, BM-MNCs of selected non-CHIP (n = 3) and CHIP (n = 3) individuals as well as LR-MDS (n = 14) and HR-MDS (n = 5) patients were FACS-sorted for hematopoietic stem and progenitor cells (HSPCs), monocytes, monocytic and polymorphonuclear myeloid-derived suppressor cells (MDSCs) and lymphocytes. The expression of IL1B, IL18, S100A9, NLRP3, PYCARD, CASP1 and NLRC4 was assessed in each population by qRT-PCR. Finally, colony forming assays were used to assess the effect of the IL-1β-neutralizing antibody canakinumab (100 µg/ml) on the hematopoietic potential of healthy CD34+ HSPCs pre-cultured with CD14+ monocytes from patients with SF3B1-mutated (n = 3) or del(5q) (n = 3) LR-MDS. Results The expression of inflammasome-related genes in the bone marrow increased from non-CHIP through CHIP to LR-MDS individuals, with overall levels of inflammasome-related gene expression being 40% higher in LR-MDS than in non-CHIP controls (p = 0.002). The precise signature of inflammation gene expression varied widely within LR-MDS patients. A PCA identified two separable inflammation states with low (cluster 1, n = 24) and high (cluster 2, n = 23) inflammation patterns (Figure 1), with CASP1, PYCARD and NLRC4 contributing most to the PCA dimensions. The clusters were associated with distinct clinical features and disease genotypes: more than 80% of the SF3B1 mutations were found in cluster 1, while all del(5q) cases were in cluster 2 (Figure 1). Compared to del(5q) samples, those with SF3B1-mutation showed lower expression of IL-1β at both mRNA (3.9-fold decrease, p = 0.0005) and protein level (8.3-fold decrease, p = 0.03). Targeted gene expression analysis of sorted cell populations revealed distinct contributions to the differential patterns of inflammation. IL1B, NLRP3, NLRC4 and S100A9 were expressed in all disease states predominantly by monocytes and monocytic MDSCs, consistent with a dominant role for monocytic cells in determining the inflammatory bone marrow environment. However, IL18 mRNA levels were highest in HSPCs. Compared to untreated controls, canakinumab augmented the colony forming activity of healthy HSPCs pre-cultured with monocytes from SF3B1-mutated MDS (+70%, p = 0.002). The effect in co-cultures with del(5q) MDS-derived monocytes was weaker (+56%, p = 0.188), showing that canakinumab efficacy in this assay does not depend on IL-1β expression level alone. Conclusion We conclude that inflammation patterns in LR-MDS are heterogeneous, highlighting the need for personalization of targeted anti-inflammatory therapies currently entering clinical trials. Our analysis suggests that high resolution studies of gene expression in specific populations or single cells will further resolve a spectrum of inflammation states relevant to prognosis and personalized therapy. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal