SF3B1 is the most prevalent splicing factor mutated in myeloid neoplasms, detected in 20-30% of patients with myelodysplasia (MDS) and associated with a milder disease phenotype. SF3B1 mutations are typically single nucleotide variations occurring in hotspots in the HEAT domain. Disease-specific SF3B1 hotspot predilection is observed; for example, K700E is most common in MDS and R625 in uveal melanoma. The K666 hotspot has been associated with increased risk of MDS disease progression to acute myeloid leukemia (AML). However, the prognostic impact across all myeloid neoplasms inclusive of myeloproliferative neoplasms (MPN) and MDS/MPN overlap syndromes has not been established. Mutation information was collated from available datasets inclusive of unpublished clinical trial cohorts across patients with myeloid neoplasms (MDS, MPN, MDS/MPN, AML) and patients with solid malignancies and clonal hematopoiesis (CH). SF3B1 hotspot mutations were correlated with additional somatic mutations, disease type and other clinical parameters. Overall, 11,744 patients with myeloid neoplasms; MDS (n=4275), AML (n=3526), MDS/MPN (n=192), MPN (n=3751) and 24,146 patients with solid malignancies and CH were included in this analysis. SF3B1 mutations were present in 11.1% of patients; 21.3% MDS, 4.1% AML, 51% MDS/MPN and 3.7% MPN. In MDS, SF3B1 mutations were predominantly present in low risk MDS (23.8%) as compared with 4.1% in intermediate to high risk MDS. This contrasts with MPN where SF3B1 mutations were present at a higher frequency in more advanced MPN (5.5% in treatment-resistant ET, 8.8% in all MF cases and further augmented at 14% in JAK inhibitor resistant high-risk thrombocytopenic MF) as compared with earlier phase MPN (2.1% and 1.3% in ET and PV respectively). In MPN SF3B1-mutant cases, JAK-STAT signaling driver mutation frequency and distribution were in line with published data; 65.6% JAK2V617F-mutated, 18.8% CALR-mutated, 8.5% MPL-mutated and 7% triple negative. As previously described, SF3B1 K700E hotspot mutations were most prevalent (frequency 49.1%), followed by K666, H622 and R625 hotspot mutations at 20.6%, 7.6% and 6.1% respectively. K700E was enriched as expected in MDS cohorts and K666 hotspot mutations were enriched in AML (42%). Although less common in MDS (11.6 - 17.8%), the frequency of K666 mutation correlated with higher IPSS-M risk category in MDS (p=0.001; chi-squared test). In contrast to MDS, SF3B1 K666 was the dominant hotspot in MPN ranging 46-70% in frequency across subtypes from earlier to more advanced MPN such as MF (Fig 1A). SF3B1 K700E mutations were more often co-mutated with epigenetic mutations such as in DNMT3A, TET2, EZH2 and KMT2C genes (Fig 1B) whereas SF3B1 K666 mutations were significantly co-mutated with NPM1 and FLT3 in AML cases (odds ratio, OR, 25.6 and 4.2 respectively, p-value <0.0001) and JAK2V617F in MPN cases (OR 3.2, p<0.0001). SF3B1 K666 mutations also showed a trend towards significant co-occurrence with adverse prognostic mutations in AML; PHF6, PTPN11 and RUNX1. No differences were observed for SF3B1 hotspots for CALR or MPL-mutated MPN but numbers were too few to draw conclusions. In patients with SF3B1 mutations, 58% were male and K666 hotspot was more frequent in males (21% as compared with 11% in females, p<0.001, chi-squared test). There were no hotspot differences observed in age, ethnicity, bloods counts or karyotype where data was available for this analysis. In CH, SF3B1 mutations were detected in 0.5% with K700E (34.2%) and K666 (26.7%) frequencies equivalent (p=0.24). In summary, this large multi-cohort myeloid neoplasm analysis reveals myeloid disease-specific SF3B1 hotspot propensity. The K666 hotspot was frequent in MPN, particularly in persons with advanced disease where SF3B1 mutations are more prevalent than earlier phase MPN. We conclude that different SF3B1 hotspots are associated with distinct clinical phenotypes and in chronic myeloid malignancies; K666 SF3B1 mutation is associated with adverse clinical course. Additional studies to elucidate the distinct biological effects of different SF3B1 mutations are warranted.
Read full abstract