The prognostic impact of mutations in spliceosomal genes for myelodysplastic syndrome patients without ring sideroblasts

Mutations in genes that are part of the splicing machinery for myelodysplastic syndromes (MDS), including MDS without ring sideroblasts (RS), have been widely investigated. The effects of these mutations on clinical outcomes have been diverse and contrasting. We examined a cohort of 129 de novo MDS patients, who did not harbor RS, for mutations affecting three spliceosomal genes (SF3B1, U2AF1, and SRSF2). The mutation rates of SF3B1, U2AF1, and SRSF2 were 7.0 %, 7.8 %, and 10.1 %, respectively. Compared with previously reported results, these rates were relatively infrequent. The SRSF2 mutation strongly correlated with old age (P < 0.001), while the mutation status of SF3B1 did not affect overall survival (OS), progression-free survival (PFS), or acute myeloid leukemia (AML) transformation. In contrast, MDS patients with mutations in U2AF1 or SRSF2 exhibited inferior PFS. The U2AF1 mutation was associated with inferior OS in low-risk MDS patients (P = 0.035). The SRSF2 mutation was somewhat associated with AML transformation (P = 0.083). Our findings suggest that the frequencies of the SF3B1, U2AF1, and SRSF2 splicing gene mutations in MDS without RS were relatively low. We also demonstrated that the U2AF1 and SRSF2 mutations were associated with an unfavorable prognostic impact in MDS patients without RS.


Background
The myelodysplastic syndromes (MDS) represent myeloid clonal hemopathies, with a relatively heterogeneous spectrum of presentation. The major clinical problems of these disorders are morbidities caused by cytopenias and the potential for MDS to evolve into acute myeloid leukemia (AML) [1]. Although cytopenias represent the major clinical challenge in low-risk disease, transformation to AML is observed in a significant number of high-risk MDS patients.
The broad range of individual genes affected by mutations indicates that a variety of molecular mechanisms are involved in the pathogenesis of MDS [2]. A number of gene mutations and cytogenetic changes have been implicated in the pathogenesis of MDS, including mutations in RAS, TP53, and RUNX1. However, mutations in these genes do not fully explain the pathogenesis of MDS as these mutations are also commonly found in other myeloid malignancies. In addition, approximately 20 % of MDS cases are not associated with any genetic changes. The genetic alterations responsible for dysplastic phenotypes and ineffective hematopoiesis of myelodysplasia are poorly understood [3].
A previous report by Murati et al. [4] described that mutations in components of the spliceosome, which are mutually exclusive, lead to splicing defects, including exon skipping, intron retention, and the use of incorrect splice sites. The consequence of mutations in spliceosomal genes is the accumulation of unspliced transcripts that affect a specific subset of mRNAs. According to Yoshida et al. [3] and Makishima et al. [5], mutations affecting spliceosomal genes that result in defective splicing could belong to a new leukemogenic pathway, and these mutations might constitute diagnostic biomarkers that could serve as therapeutic targets.
A recent study by Damm et al. [2] revealed that splice gene mutations are among the most frequent molecular aberrations in MDS. They might define distinct clinical phenotypes and show preferential association for mutations targeting transcriptional regulation. These genotype-phenotype associations have been demonstrated for somatic spliceosomal gene mutations in MDS with ring sideroblasts (RS). Although there have been a number of studies investigating spliceosomal mutations in MDS without RS, the effects of these mutations on clinical outcomes have not been uniform.
We investigated the prevalence and clinical impact of mutations in splicing factor 3 subunit b1 (SF3B1), U2 small nuclear RNA auxiliary factor 1 (U2AF1), and serine arginine-rich splicing factor 2 (SRSF2) among a cohort of MDS patients without RS.

Mutation analyses of spliceosomal genes
Genomic DNA from each MDS patient was extracted using the AccuPrep Genomic DNA Extraction Kit (Bioneer, Daejeon, Korea) according to the manufacturer's instructions. The detection of mutations in SF3B1, U2AF1, and SRSF2 was conducted using polymerase chain reaction (PCR) followed by direct sequencing. For direct sequencing of the spliceosomal genes, six primer pairs were used (Additional file 1: Table S1) according to a published protocol (Additional file 2), with some minor modifications. Gene sequences were compared using Blast2 (http:// blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE_TYPE=BlastSearch& BLAST_SPEC=blast2seq&LINK_LOC=align2seq) to obtain preliminary evidence regarding polymorphisms, mutations, and for translation of amino acids. Results obtained from MDS patients were confirmed on an online database (http:// genewindow.nci.nih.gov/Welcome; Additional file 2). The aberrant status of SF3B1, U2AF1, and SRSF2, was confirmed by TA cloning (Fig. 1) using the pGEM-T Easy vector (Promega, Madison, WI, USA). For each spliceosomal gene, three MDS patients representative of the typical heterozygous form of the gene were selected (Additional file 2).

Cytogenetic analysis
Chromosomal analysis (G-banding) was performed on preparations from 48-h bone marrow cell cultures where mitogens were not added, according to a protocol from the American Type Culture Collection. Aberrations in chromosomes were described according to the international system for cytogenetic nomenclature 2005 and 2009.

Statistical analyses
The χ 2 test or Fisher's exact test was performed to determine the significance of associations between SF3B1, U2AF1, and SRSF2 mutations and other parameters, including sex, WHO classification, karyotypes, and IPSS-R risk classification. Student's t-test was used to compare continuous variables such as age and hemograms. Kaplan-Meier estimation was used to plot survival curves, and logrank tests were used to calculate the difference between survival curves. Cox proportional hazard regression analysis was used to dissect the individual impact of prognostic factors for overall survival (OS), progression-free survival (PFS), and acute myeloid leukemia (AML) transformation. All tests were two-tailed, and a P-value of less than 0.05

Results
Mutation status of SF3B1, U2AF1, and SRSF2 in MDS patients Mutations in one of the spliceosomal genes (SF3B1, U2AF1, and SRSF2) were observed in 24.8 % (32/129) of MDS patients . Among the 129 MDS patients, nine were identified as harboring a mutation in SF3B1 (7.0 %), 10 patients had mutations in U2AF1 (7.8 %), and 13 patients exhibited a mutation in SRSF2 (10.1 %). All 129 MDS patients in this study were without RS. The SF3B1, U2AF1, and SRSF2 mutations were mutually exclusive, with none of the patients having more than one of these genes affected (Tables 1 and 2). The mutations in SF3B1, U2AF1, and SRSF2 were all heterozygous point mutations (n = 32; Table 2). The aberrant status of SF3B1, U2AF1, and SRSF2 was confirmed by TA cloning and direct sequencing (Fig. 1).
Patient characteristics with respect to SF3B1, U2AF1, and SRSF2 mutation status The clinical and hematological characteristics of patients with mutated (mut) versus wild-type (wt) SF3B1, U2AF1, and SRSF2 are shown in Table 1. Patients with SF3B1 mutations showed significant differences in karyotype (P = 0.013). Positive cytogenetic findings, such as normal karyotype, −Y only, del(5q) alone, and del(20q) alone were more frequent in SF3B1 mut patients than in SF3B1 wt patients (88.9 vs. 75.8 %). Poor cytogenetic findings, such as complex karyotype, and abnormalities in chromosome 7 were more apparent in SF3B1 wt patients than in SF3B1 mut patients (10.0 vs. 0 %). There were no significant differences in age, sex, blood counts, bone marrow blasts, WHO subtype, and IPSS-R risk classification between SF3B1 mut and SF3B1 wt patients. Nevertheless, lower risk MDS patients, such as those with refractory cytopenia with unilineage dysplasia (RCUD) or refractory cytopenia with multilineage dysplasia (RCMD), were represented in higher proportions among SF3B1 mut patients than SF3B1 wt patients (66.7 vs. 57.5 %). For higher risk MDS patients, such as those with refractory anemia with excess blasts-1 (RAEB-1) or RAEB-2, there was a lower proportion of SF3B1 mut patients than SF3B1 wt patients (22.2 vs. 36.7 %). Patients harboring mutations in U2AF1 were mainly male (8/10) and exhibited lower hemoglobin levels (mean: 8.4 vs. 9.7 g/dL for U2AF1 mut vs. U2AF1 wt ; P = 0.063). Our cytogenetic results revealed meaningful differences between U2AF1 mut and U2AF1 wt patients (P = 0.022). Positive cytogenetic findings were more frequently observed for U2AF1 wt than U2AF1 mut patients (78.3 vs. 60.0 %), while poor cytogenetic findings were more common in U2AF1 mut patients (10.0 vs. 9.2 %). In contrast, no significant differences were identified between U2AF1 mut and U2AF1 wt patients for age, sex, blood counts, bone marrow blasts, WHO subtype, and IPSS-R risk classification. The higher risk MDS patients (RAEB-1 or RAEB-2) were more likely to be U2AF1 mut patients (60.0 vs. 33.6 %), while lower risk MDS patients (RCUD or RCMD) were less likely to be U2AF1 mut individuals (40.0 vs. 59.7 %) (P = 0.629).
The SRSF2 mut patients were older than SRSF2 wt patients (mean: 71.5 vs. 62.8 years; P < 0.001) and mostly male (9/13). Similar to the U2AF1 mut patients, those with SRSF2 mutations displayed a significant difference in cytogenetic results (P = 0.048). Good cytogenetic findings were more frequently seen for SRSF2 wt patients (79.4 vs. 53.8 % in SRSF2 mut patients), while poor cytogenetic findings were more common for SRSF2 mut patients (23.1 vs. 7.8 % in SRSF2 wt patients). We observed no significant differences in sex, blood counts, bone marrow blasts, WHO subtype, and IPSS-R risk classification between SRSF2 mut and SRSF2 wt patients. The higher risk MDS patients (RAEB-1 or RAEB-2) were more likely to be SRSF2 mut patients (46.2 vs. 34.5 %), while lower risk MDS patients (RCUD or RCMD) were less likely to be SRSF2 mut patients (53.9 vs. 58.6 %) (P = 0.094).
The IPSS-R was used to derive clinical prognosis for MDS. To establish whether the mutation status of spliceosomal genes can add to the predictive power of IPSS-R, We evaluated OS, PFS, and AML probabilities according to the mutation status of spliceosomal genes in all MDS patients (Fig. 2a-i), and subgroups of MDS patients (Fig. 3a-d), using Kaplan-Meier estimation. No differences in survival were seen for all MDS patients with or without mutations in SF3B1 (Fig. 2a, d, and g). Patients carrying a mutation in U2AF1 (P = 0.009; Fig. 2e) or SRSF2 (P = 0.001; Fig. 2f) exhibited significantly lower PFS compared with wild-types. The presence of a SRSF2 mutation was a somewhat unfavorable prognostic factor for AML transformation (P = 0.054; Fig. 2i).
MDS subgroup analysis revealed that the poor impact of a U2AF1 mutation on OS was only demonstrated in the lower risk groups (very low and low) defined by IPSS-R (P = 0.035; Fig. 3a). In addition, patients harboring the U2AF1 mutation showed inferior PFS in the higher risk groups (RAEB-1 or RAEB-2) defined by WHO 2008 criteria (P = 0.045; Fig. 3b). Patients with the SRSF2 mutation showed inferior PFS in the lower risk groups (RCUD or RCMD) defined by WHO 2008 criteria (P = 0.004; Fig. 3c). Patients with a SRSF2 mutation exhibited a somewhat increased rate for AML transformation among lower risk (RCUD or RCMD) MDS patients (P = 0.083; Fig. 3d). No survival differences were seen between MDS patients with or without the SF3B1 mutation (data not shown).

Discussion
Recent reports regarding whole exome sequencing in MDS patients by Yoshida et al. [3] and Papaemmanuil et al. [7] suggest that spliceosome mutations have some clinical relevance. Identifying the impact of these mutations on MDS pathogenesis holds some promise for the therapeutic modulation of mRNA splicing [8]. The exact functional consequences of these spliceosomal mutations in MDS pathogenesis and other hematological malignancies remain largely unknown, and are being intensely investigated [9]. The molecular diversity of MDS corresponds to the clinical and phenotypic heterogeneities of these syndromes. Moreover, molecular defects could potentially serve as biomarkers for the identification of therapeutic targets [5]. To date, these genotype-phenotype associations of  For SF3B1 mut patients, no AML transformation was seen AML, acute myeloid leukemia; CI, confidence interval; HR, hazard ratio; IPSS-R, revised International Prognostic Scoring System; mut, mutated; WT, wild-type MDS have been described in many previous studies. Numerous researchers have investigated spliceosomal mutations in MDS without RS; however, the effects of these mutations on clinical outcomes have not been uniform. We investigated the prevalence and prognostic implication of the SF3B1, U2AF1, and SRSF2 mutations in MDS patients without RS from Korea.
Our findings indicate that the SF3B1, U2AF1, and SRSF2 mutations were relatively infrequent in MDS patients without RS, contradicting the results from a previous study. In addition, our results demonstrate that the U2AF1 and SRSF2 mutations, unlike SF3B1, were associated with a negative prognostic impact for MDS patients without RS.
Spliceosomes in the nucleus are complexes composed of small nuclear RNAs (snRNA) and numerous protein subunits. These spliceosomes serve to remove introns from genes that encode proteins [10]. Identifying these genes and understanding the mechanisms involved in aberrant splicing might lead to advancements in diagnosis and treatment of MDS and other diseases [11]. According to a recent report by Makishima et al., mutations affecting , and probability of AML transformation g-i for the total MDS patient cohort (n = 129), stratified according to SF3B1, U2AF1, and SRSF2 mutation status. AML, acute myeloid leukemia; wt, wild-type; mut, mutant spliceosomal genes that result in defective splicing belong to a new leukemogenic pathway, with these mutations possibly constituting diagnostic biomarkers that could be therapeutic targets [5].
These spliceosomal gene mutations occur at varying frequencies for different disease subtypes, and contribute to differences in survival outcomes [9]. The SF3B1 gene is located on chromosome 2q33.1 and codes for the SF3B1 protein complex, which is involved in the early stages of spliceosome assembly. U2AF1 gene is located on chromosome 21q22, and encodes proteins that play a role in the early steps of 3′ splice site recognition. The SRSF2 gene is located on chromosome 17q25.2, with the coding protein known to play a role in preventing exon skipping and ensuring the accuracy of splicing [12].
It was previously reported that the incidence of MDS with RS is far less common than that of MDS without RS in the Korean population [13,14]. Consistent with Overall survival a and progression-free survival b-c are affected by U2AF1 or SRSF2 genotypes according to subgroup analysis of MDS patients. The probability of AML progression was increased for RCUD and RCMD patients with a mutation in SRSF2 d. AML, acute myeloid leukemia; IPSS-R, revised International Prognostic Scoring System; mut, mutant; RAEB, refractory anemia with excess blasts; RCMD, refractory cytopenia with multilineage dysplasia; RCUD, refractory cytopenia with unilineage dysplasia; wt, wild-type previous studies, our study population comprised 129 MDS patients without RS. For this cohort, the mutation rates of SF3B1, U2AF1, and SRSF2 were 7.0, 7.8, and 10.1 %, respectively ( Table 1). The occurrence of mutations in these genes, for MDS patients without RS, were relatively infrequent compared with that seen in earlier studies [3,8,9,14]. Hahn and Scott reported that the p.Lys700Glu was the most recurrently occurring alteration in both MDS and chronic lymphocytic leukemia [10]. In the current study, this particular mutation was the most common seen in spliceosomal genes likewise (Table 2).
Malcovati et al. reported that only 5.3 % (2/38) of patients with AML evolving from MDS carried a somatic mutation in SF3B1 [15]. In our current study, none of the SF3B1 mut MDS patients progressed into AML, and these patients were more likely to present with advantageous cytogenetic findings. However, U2AF1 mut and SRSF2 mut patients were considered to belong to higher risk MDS groups or to have a poor cytogenetic findings (Table 1).
We also found that the U2AF1 mutation mainly occurred in males and correlated with relatively low hemoglobin levels. It was previously that mutations in U2AF1 confer the suppression of growth in vitro [3], possibly contributing to the cytopenias seen in U2AF1 mut patients within the current MDS cohort. Occurrence of the SRSF2 mutation strongly correlated with older individuals (P < 0.001), similar to the findings of Wu et al. [16] (Table 1).
We found that the IPSS-R total score had a strong association with OS, PFS, and AML transformation (Table 4). However, the prognostic impact of spliceosome gene mutations in MDS patients remains controversial [16]. Some studies have reported that SF3B1 mutations are a marker of favorable outcomes for MDS [7,15]. However, results from other studies [17], including our analysis in the current study, indicate that SF3B1 mutations do not represent an independent prognostic factor (Tables 3 and 4, Fig. 1). These differences could be attributed to the heterogeneity of the disease itself, the composition of patient populations, and the various treatment strategies used [17,18].
Regarding the U2AF1 mutation, results from one study concluded that it did not influence OS [19], while another report claimed that it was associated with shorter OS [5]. Analysis of our whole cohort, or even subgroup analysis of MDS patients, revealed inferior OS and PFS for U2AF1 mut patients (Figs. 2e and 3a-b). This negative prognostic impact for PFS was also seen when we conducted univariate or multivariate Cox regression analysis (Tables 3 and 4), further supporting the idea that the U2AF1 mutation could be an independent prognostic marker for MDS.
The SRSF2 mutation negatively affected PFS in MDS patients, especially for those in the lower risk MDS groups (Figs. 2f and 3c). We also found that the SRSF2 mutation was an independent prognostic factor for a poor PFS outcome (Tables 3 and 4). Consistent with findings by Thol et al., who reported that SRSF2 mutations were associated with an increased risk of progression to AML [18], we observed a somewhat significant impact of the SRSF2 mutation on the progression time to AML transformation (Figs. 2i and 3d). In a previous study, deletion of SRSF2 contributes to genomic instability, which is a predictive marker for adverse outcomes in MDS, and possibly explains why SRSF2 mutations confer a strong adverse effect [18].

Conclusions
In summary, we observed that mutations in SF3B1, U2AF1, and SRSF2, in MDS patients without RS, were relatively infrequent molecular events. The mutation status of SF3B1 was not associated with OS, PFS, or AML transformation, regardless of the groupings used in our analyses. In contrast, all U2AF1 mut and SRSF2 mut patients displayed inferior PFS. We observed that mutations in U2AF1 were associated with inferior OS in the lower risk MDS groups defined by IPSS-R (very low or low risk) and that there was somewhat of an association between AML transformation and mutations in SRSF2.