P697: distinct splicing alterations associated with clinical response to luspatercept in patients with lower‑risk myelodysplastic syndromes from the medalist study

Background: Patients with erythropoiesis-stimulating agent (ESA)-refractory, transfusion-dependent (TD) lower-risk myelodysplastic syndromes (LR-MDS) have limited treatment options and poor clinical outcomes. Luspatercept is an erythroid maturation agent that sequesters certain TGFβ superfamily ligands leading to downregulation of SMAD2/3 signaling and an increase in availability of erythroid transcription factor GATA1. In the phase 3 MEDALIST study, luspatercept led to red blood cell (RBC) transfusion independence (RBC-TI) ≥8 weeks in 38% of patients with LR-MDS in the first 24 weeks. Luspatercept is approved to treat anemia in patients with LR-MDS with ring sideroblasts who require RBC transfusions and are ESA-refractory. Aims: Here, a multi-omics biomarker analysis was performed to identify biological associations of luspatercept response and non-response in MEDALIST patients. Methods: MEDALIST (NCT02631070) was a randomized, phase 3 study of luspatercept vs placebo in patients with LR-MDS. RNA sequencing analysis, using 75 base pair end reads, was performed on bone marrow (BM) samples collected at baseline and 25 weeks posttreatment from luspatercept responders (R; n=14; defined as achieving RBC-TI ≥8 weeks during weeks 1–24) and non-responders (NR; n=9). Proteomics were performed on BM samples using trapped ion mobility mass spectrometry. Results: Conventional cytogenetics and overall mutational status were not associated with response to luspatercept. Global transcriptomic analysis on baseline BM samples yielded 49 differentially expressed genes in R and NR; significant differences in splice isoforms were observed. In total, 2073 differential splicing events were identified, with the majority of aberrant splicing observed in NR (2659 in NR vs 586 in R). Skipped exons were predominantly splicing alterations; a total of 1698 differential events were observed in NR vs 238 in R. Pathways affecting RNA splicing were the predominant group affected by differential exon usage. Production of a shorter, differentially spliced form of GATA1 due to exon 2 skipping was significantly associated with response to luspatercept. R had a higher short:long GATA1 isoform ratio vs NR (12% vs 6%; P=0.039). A 15bp intronic retention event next to the terminal exon of GATA1 was confirmed to be associated with SF3B1-mutated MDS in an independent cohort of patient samples and in a cell-based model. This intronic insertional event, leading to the production of altered GATA1, may act in a dominant negative manner, suggestive of compromised GATA1 transcriptional activity in SF3B1-mutated LR-MDS. To validate these findings, the same BM samples were analyzed by mass spectrometry. Functional enrichment analysis of BM proteomic data showed upregulation of protein in RNA splicing-related pathways at baseline in NR vs R, including the KEGG database spliceosome pathway (P=0.009; normalized enrichment score [NES]=1.536). Summary/Conclusion: These results demonstrate a significant association of increased splicing events, which in part predict reduced response to luspatercept treatment as a second line in ESA-refractory TD LR-MDS, suggesting a complex intrinsic cell abnormality that cannot be rescued by a single agent. Response to luspatercept was associated with differential splicing of GATA1, leading to a higher proportion of a shorter, inactive isoform. Since luspatercept has been shown to increase GATA1-related erythroid gene expression, indicating that patients with LR-MDS-associated anemia driven by expression of a shorter GATA1 isoform may respond to luspatercept. Keywords: Myelodysplastic syndrome, Anemia, Transcription factor, Alternative splicing