Splicing factor mutations in AML

In this issue of Blood, Taskesen et al propose a novel subset of myeloid neoplasms that encompasses splicing factor (SF) mutation-positive refractory anemia with excess blasts (RAEB) and acute myeloid leukemia (AML), which is composed of 2 molecularly and clinically distinct subgroups.¹ 

Myeloid neoplasms comprise a wide variety of clinically and pathologically distinct entities commonly characterized by neoplastic proliferation of myeloid lineage cells with deregulated differentiation.²  Although straightforward for typical cases, distinction among different subtypes of myeloid neoplasms may not be unambiguous, complicating the diagnosis of and the therapeutic choice in borderline cases. According to the current World Health Organization (WHO) classification system, for example, distinction between AML with low blast counts (AML-LBC) and the RAEB subtype of myelodysplastic syndromes (MDS) is made only by the presence or absence of ≥20% bone marrow blasts.²  However, these criteria, based on the blast counts, would lead to a diagnosis of AML for some patients who might be biologically/clinically more properly classified as RAEB or vice versa. In this regard, during the past 10 years, our knowledge about the molecular pathogenesis of myeloid neoplasms has been substantially improved by the identification of major driver mutations causally related to these neoplasms,^3-6  which may help better understanding, classification, diagnosis, and management of different entities of myeloid neoplasms.

Taskesen et al focused on pre–messenger RNA SF mutations and tested their hypothesis that SF mutations could define a novel subset of myeloid neoplasm that includes SF mutation–positive RAEB and AML as well as AML-LBC. SF mutations are a novel class of driver mutations recently identified through whole exome sequencing of myeloid dysplasia.^7,8  Affecting at least 8 different components of the pre–messenger RNA splicing machinery, including SF3B1, SRSF2, U2AF35, ZRSR2, U2AF65, SF1, SF3A1, and PRPF40B, SF mutations represent among the most frequent genetic lesions in MDS (45% to 85% of MDS depending on their subtypes) and other related myeloid neoplasms with myelodysplasia but much less common (5% to 10%) in de novo AML and classical myeloproliferative neoplasms.^3,7  In the current study, authors investigated 47 RAEB, 29 AML-LBC, and 325 other AML patients for predominant hot-spot mutations in SF3B1, U2AF35, and SRSF2, and found that patients with RAEB and AML-LBC showed significantly higher SF mutation rates compared with other AML and, except for white cell counts and bone marrow blast counts, shared a highly similar clinical, cytological, and molecular profile. Also, SF-mutated patients among RAEB, AML-LBC, and other AML categories had similar clinical phenotypes, including lower blast counts, older age, lower white cell counts, and higher erythroblast counts in bone marrow compared with SF-unmuted cases, indicating that SF-mutated cases comprised a distinct entity among MDS/AML. In accordance with this, a hierarchical clustering of AML/AML-LBC/RAEB cases based on combined, but not separate, gene expression (GEP) and DNA-methylation profiles (DMP) identified 2 distinct clusters (#3 and #11) highly enriched for SF mutations and RAEB/AML-LBC phenotypes with significantly lower bone marrow blast counts, together with known clusters characterized by the presence of t(8;21), inv(16), t(15;17), and CEBPA mutations.^9,10  Interestingly, whole exome sequencing of 14 cases with SF mutations within the 2 clusters revealed mutations in 3 genes implicated in RNA splicing. Among the 2 clusters, #11 was characterized by an erythroid signature with higher erythroblast percentages and differentially expressed or hypomethylated genes involved in erythroid development, whereas cluster #3 was significantly enriched for NRAS/KRAS mutations and poor overall survivals compared with other patients.

Although the conclusions need to be validated further in independent studies with more comprehensive detection of SF mutations, not just hot-spot mutations, the present study points to an intriguing possibility that SF mutations could override the conventional separation between AML and MDS and help to define novel biological subtypes of myeloid malignancies for better understanding and management of AML/MDS. However, the biological basis for the SF-mutated AML is still unclear, as is the reason why the 2 SF-mutated clusters are identified only through combined GEP and DMP analysis. Finally, and more importantly, the impact of the SF-mutated AML or the novel clusters identified through GEP/DMP profiling on the choice of therapies and patients’ outcomes should be clarified before these subtypes are found to be relevant to clinical practice.