Somatic SF3B1 mutation in myelodysplasia with ring sideroblasts

Abstract

Background: Myelodysplastic syndromes are a diverse and common group of chronic hematologic cancers. The identification of new genetic lesions could facilitate new diagnostic and therapeutic strategies.

Methods: We used massively parallel sequencing technology to identify somatically acquired point mutations across all protein-coding exons in the genome in 9 patients with low-grade myelodysplasia. Targeted resequencing of the gene encoding RNA splicing factor 3B, subunit 1 (SF3B1), was also performed in a cohort of 2087 patients with myeloid or other cancers.

Results: We identified 64 point mutations in the 9 patients. Recurrent somatically acquired mutations were identified in SF3B1. Follow-up revealed SF3B1 mutations in 72 of 354 patients (20%) with myelodysplastic syndromes, with particularly high frequency among patients whose disease was characterized by ring sideroblasts (53 of 82 [65%]). The gene was also mutated in 1 to 5% of patients with a variety of other tumor types. The observed mutations were less deleterious than was expected on the basis of chance, suggesting that the mutated protein retains structural integrity with altered function. SF3B1 mutations were associated with down-regulation of key gene networks, including core mitochondrial pathways. Clinically, patients with SF3B1 mutations had fewer cytopenias and longer event-free survival than patients without SF3B1 mutations.

Conclusions: Mutations in SF3B1 implicate abnormalities of messenger RNA splicing in the pathogenesis of myelodysplastic syndromes. (Funded by the Wellcome Trust and others.).

Figures

Figure 1. Exome Sequencing in Nine Patients with Low-Grade Myelodysplastic Syndromes (MDS)

Panel A shows the distribution of numbers and categories of somatically acquired point mutations among the nine patients. No mutations were found for Patient 5. Panel B shows the mutation spectrum for somatically acquired point mutations. Panel C shows the fraction of reads reporting mutated alleles from exome-sequencing data for each patient. No mutations were found for Patient 5. Mutations in known MDS genes or recurrently mutated genes identified in this screen are shown as colored points, with nonrecurrent mutations as gray points. P values were calculated with the use of chi-square tests of heterogeneity in observed allelic ratios for mutations in patients with more than two mutations.

Figure 2. Distribution of Missense Mutations in SF3B1

The SF3B1 gene encodes a protein with an N-terminal domain together with a C-terminal region consisting of 22 so-called HEAT domains (Huntingtin, elongation factor 3, protein phosphatase 2A, and the yeast PI3-kinase TOR1). Two mutations (R1041H and E491G) fell outside the second through the eighth HEAT domains and are not shown. AML denotes acute myeloid lymphoma, CLL chronic lymphocytic leukemia, CML chronic myeloid leukemia, CMML chronic myelomonocytic leukemia, ET essential thrombocythemia, PMF primary myelofibrosis, RA refractory anemia, RAEB refractory anemia with excess blasts, RARS refractory anemia with ring sideroblasts, RCMD refractory cytopenia with multilineage dysplasia, and RCMD-RS RCMD and ring sideroblasts.

Figure 3. Modeling of Mutations and Results of Clinical Studies

Panel A shows bars representing the mean scores for computer-generated random missense mutations. The height of the bars represents the frequency histogram of this null distribution. The red squares indicate the mean scores for the somatic mutations on exons 12 through 15 that could be scored, with higher scores representing effects predicted to be less deleterious than those of random mutations. The somatic set has a significantly higher mean score than the null set for SF3B1 mutations (P<0.001). This is in contrast to the tumor-suppressor genes PTEN (encoding the phosphatase and tensin homologue) and NF1 (encoding neurofibromin 1), where observed mutations have a significantly lower score than the null set. Panel B shows a volcano plot of the distribution of normalized enrichment scores for various gene sets according to their nominal P values. Negative enrichment scores indicate gene sets that are down-regulated in patients with an SF3B1 mutation as compared with patients without such a mutation. Gene sets with significant enrichment and a false discovery rate (FDR) of less than 10% are shown in green. Panel C shows enrichment plots for four mitochondrial gene sets. Each vertical stripe represents the rank of a gene within the gene set, among all 19,578 genes represented on the microarray. A low rank indicates genes that are especially downregulated among patients with SF3B1 mutations as compared with patients without such mutations, after adjustment for disease type. The name of the gene set is shown to the left, and the q value (the minimum false discovery rate at which the test may be called significant) is shown to the right. Panel D shows box-andwhisker plots for hemoglobin levels, white-cell counts, and platelet counts in patients with myelodysplastic syndromes according to the presence or absence of SF3B1 mutations. The central horizontal line within each box indicates the median, with the top and bottom edges of each box indicating the interquartile range. The I bars extend to 1.5 times the interquartile range, with the circles indicating outlier data for individual patients. Panel E shows Kaplan–Meier curves for event-free survival during the study period among patients with myelodysplastic syndromes, according to the presence or absence of SF3B1 mutations.