DECIPHERING THE ROLE OF IMMUNE SYSTEM DYSFUNCTION IN MYELODYSPLASTIC SYNDROMES CLASSIFICATION AND PROGNOSIS

Introduction: Myelodysplastic syndromes (MDS) are a heterogeneous group of myeloid neoplasms with variable clinical outcomes and an increased risk of progression to acute myeloid leukemia (AML). Given this heterogeneity, the use of classifications and scoring systems is of fundamental importance to identify the disease subtype and to evaluate patient’s prognosis but fails in identifying patients who will respond to hypomethylating agents (HMAs), which represent the first line treatment for MDS patients so far. Several studies have underlined the role of immune dysregulation in MDS pathogenesis and progression. However, immunologic information is currently omitted from risk scores because there is still no standard method for evaluating patient’s immune status. The addition of comprehensive immunologic data to prognostic models could further help to refine risk stratification and predict therapy response. Methods: Here, we took advantage of high-dimensional flow cytometry to perform a comprehensive analysis of the immunologic landscape in bone marrow (BM) and peripheral blood (PB) of 154 MDS and AML post-MDS patients, investigating T lymphocytes, Natural Killer (NK) and Myeloid cells before and after HMA treatment. To analyze the immune cell subset distribution and phenotype, we implemented an unsupervised pipeline using Phenograph algorithm. HDBSCAN was used to clusterize MDS patients in an unsupervised manner according to their immune features. Lastly, we further characterized each immunological group by integrating bulk RNA-seq data of BM CD34+ cells isolated from patients. Results: Classical manual gating analysis revealed that in advanced stages of the disease both BM and PB displayed an immunosuppressive environment with dysfunctional T, NK and myeloid compartments that favor tumor immune escape. The unsupervised analysis identified 5 immunological groups of MDS patients characterized by different grade of immune dysfunction, related to different prognosis and response to HMA therapy. Moreover, patients classified within the same MDS subtype or prognostic risk category were subdivided into different immunological groups, underlining the importance of immunological differences to better stratify MDS patients. RNA-seq data of blast cells revealed distinct inflammatory signatures of the different immunological groups that correlated with the loss of function of immune cells. Finally, we developed a decision tree for the automatic patient’s classification within immunological groups based on four immune populations that are easy to detect with a restricted number of markers, to assign a score to the immune dysfunction and integrate this information to the existing risk stratifications. Conclusions: Taken together, our data provide evidence that the evaluation of the immune signature of MDS patients can improve MDS risk stratification and could help in predicting the response to HMA treatment as well as in identifying innovative and specific therapeutic targets.