PF210 CLINICAL ASPECTS AND DIFFERENTIAL SPLICING IN ACUTE MYELOID LEUKEMIA PATIENTS WITH SRSF2, U2AF1 AND SF3B1 MUTATIONS

Abstract

Background: Splicing factor (SF) mutations are recurrent events in hematopoietic malignancies. Their clinical characteristics and aberrant splicing patterns have been explored in myelodysplasia, however their role in acute myeloid leukaemia (AML) is not well defined. Aims: The aim of this study was the comprehensive clinical and functional analysis of AML samples with mutations in the three most commonly afflicted splicing factor genes, SRSF2, U2AF1 and SF3B1. Methods: To this end, we determined co-occurence of SF mutations and other recurrent alterations in AML and examined their prognostic role in two large independent cohorts, encompassing a total of 2442 intensively treated patients. The clinical analysis was complemented by RNA sequencing of 246 patients to quantify differential isoform expression and identify targets of splicing dysregulation. Results: As shown before, SF mutations occur more frequently in secondary AML and are associated with older age. Additionally, mutations in all three SF genes co-occur significantly with BCOR and RUNX1 mutations and display an exclusion pattern with NPM1 mutations. Splicing factor mutations associate with inferior relapse-free and overall survival. However, in multivariate models including parameters of the ELN 2017 classification, mutations in SRSF2 and SF3B1 do not represent distinct independent prognostic markers. U2AF1 mutations still have inferior relapse-free survival in the multivariate model, while patients with the specific point mutation U2AF1 (S34F) show inferior overall survival as well. In RNA sequencing, we found differential expression to be restricted to a small number of transcript isoforms, some of which displayed large fold-change differences between SF mutated and SF wildtype patients. We found little overlap of affected isoforms between patients harboring different SF mutations. Hierarchical clustering of the samples based on differential expression revealed a splicing profile highly characteristic for each individual SF mutation. Gene ontology analysis revealed several enriched pathways, including but not limited to genes involved in cell proliferation, apoptosis, and immune response. Moreover, we determined differential splice junction usage of both known and novel splice junctions. We detected genome-wide aberrant splicing which was characterized by motif-dependent, decreased splice junction utilization in SF mutated samples. The affected sites included both novel and non-canonical splice junctions. Targets of splicing dysregulation included several genes with a well-known role in cancer, such as NPM1, TET2 and WT1. Furthermore, we identified aberrant splicing patterns in most of the isoforms (78.9%) we determined previously as differentially expressed. Summary/Conclusion: In summary, these results verify the known associations of SF mutations with distinct AML markers and provide evidence of previously unknown correlations. At the same time, they suggest that while SF mutations associate with inferior survival, they are not clear independent prognostic markers. Our findings support previous studies that have characterized SF mutations as “change-of-function”, however they also introduce the possibility of splicing insufficiency as an additional pathomechanism. We conclude that SF mutations are critical events in early leukemogenesis with a large genomic impact, however the prognostic value of SRSF2 and SF3B1 mutations is lost after transformation to AML.