Abstract

Mutations in genes of the splicing machinery have been described recently in myelodysplastic syndromes (MDS). In the present study, we examined a cohort of 193 MDS patients for mutations in SRSF2, U2AF1 (synonym U2AF35), ZRSR2, and, as described previously, SF3B1, in the context of other molecular markers, including mutations in ASXL1, RUNX1, NRAS, TP53, IDH1, IDH2, NPM1, and DNMT3A. Mutations in SRSF2, U2AF1, ZRSR2, and SF3B1 were found in 24 (12.4%), 14 (7.3%), 6 (3.1%), and 28 (14.5%) patients, respectively, corresponding to a total of 67 of 193 MDS patients (34.7%). SRSF2 mutations were associated with RUNX1 (P < .001) and IDH1 (P = .013) mutations, whereas U2AF1 mutations were associated with ASXL1 (P = .005) and DNMT3A (P = .004) mutations. In univariate analysis, mutated SRSF2 predicted shorter overall survival and more frequent acute myeloid leukemia progression compared with wild-type SRSF2, whereas mutated U2AF1, ZRSR2, and SF3B1 had no impact on patient outcome. In multivariate analysis, SRSF2 remained an independent poor risk marker for overall survival (hazard ratio = 2.3; 95% confidence interval, 1.28-4.13; P = .017) and acute myeloid leukemia progression (hazard ratio = 2.83; 95% confidence interval, 1.31-6.12; P = .008). These results show a negative prognostic impact of SRSF2 mutations in MDS. SRSF2 mutations may become useful for clinical risk stratification and treatment decisions in the future.

Introduction

Myelodysplastic syndromes (MDS) are neoplastic disorders of the hematopoietic stem cell that are characterized by peripheral cytopenias and a variable propensity for transformation to acute myeloid leukemia (AML).1  The current concept of the disease considers gene mutations, deregulated gene expression, and epigenetic changes as key steps in the pathogenesis of the disease.2  Several mutations have been described to occur in MDS, including mutations in TET2, ASXL1, RUNX1, EZH2, TP53, NRAS, KRAS, CBL, ETV6, IDH1, and IDH2, with the recent addition of DNMT3A.2–4  However, these mutations do not appear to be specific for MDS because they also occur in other myeloid malignancies at variable frequencies.5  Moreover, little is known about the mechanisms leading to the disease-defining features of MDS. Recently, mutations in Splicing factor 3B subunit 1 (SF3B1) were discovered that appear to be not only specifically associated with MDS but even with the subtypes of MDS that show ring sideroblasts (RARS and RCMD-RS).6 SF3B1 encodes a protein involved in the spliceosome pathway. Simultaneously, other splicing genes such as U2 small nuclear RNA auxiliary factor 1 (U2AF1, synonym U2AF35), serine/arginine-rich splicing factor 2 (SRSF2), and U2 small nuclear ribonucleoprotein auxiliary factor 35 kDa subunit-related protein 2 (ZRSR2) involved in the spliceosome pathway have also been reported to be frequently mutated in MDS.7  However, although these gene mutations also appear to be most prevalent in MDS, they are not associated with the presence of ring sideroblasts.7  The mechanism of how mutations in genes belonging to the splicing machinery can lead to the pathogenesis of MDS is still unknown. Whereas some data suggest a positive or no prognostic impact of SF3B1 mutations,6,8–10  the prognostic impact and clinical characteristics of patients with mutations in other splicing genes such as SRSF2, U2AF1, and ZRSR2 are unknown.

In the present study, we investigated the frequency and prognostic influence of mutations in SRSF2, U2AF1, and ZRSR2 in the context of other prognostic clinical and molecular markers in a cohort of 193 MDS patients.

Methods

Patients

Cell samples from 193 MDS patients were collected at the time of enrollment in clinical trials. MDS patients were enrolled in multicenter treatment trials investigating the use of all-trans retinoic acid,11  antithymocyte globulin (NCT00004208),12,13  deferasirox,14  lenalidomide, or thalidomide for the treatment of MDS; demethylating agents were not used for this patient cohort. DNA from 50 healthy blood donors (age, 18-60 years) was obtained from the Institute of Transfusion Medicine, Hannover Medical School (Hannover, Germany). Clinical and hematologic data were recorded after MDS patients gave their informed consent in accordance with the Declaration of Helsinki, and the scientific analysis of the samples was approved by the institutional review board of Hannover Medical School (ethical vote 2467). The distribution of World Health Organization (WHO) subtypes, International Prognostic Scoring System (IPSS) risk groups, and cytogenetic risk groups is shown in Table 1 and supplemental Tables 2 and 3 (available on the Blood Web site; see the Supplemental Materials link at the top of the online article). The study cohort included more low-risk (n = 96) than high-risk (n = 51) MDS patients. Patients with chronic myelomonocytic leukemia were excluded. Follow-up information was available for 154 of the 193 MDS patients and was updated through documented clinic visits as well as telephone calls to patients, their doctors, or local registry offices.

Table 1

Comparison of clinical and molecular characteristics of 193 MDS patients according to SRSF2 mutation status

Characteristic SRSF2 mutated (n = 24) SRSF2 WT (n = 169) P 
Age, y   .12 
    Median 69 65  
    Range 56-83 36-92  
Sex, n (%)   .059 
    Male 19 (79) 100  
    Female 5 (21) 69  
WHO subtype, n (%)   .86 
    RA 7 (29) 31 (18)  
    RARS 2 (8) 18 (11)  
    del (5q) 1 (4) 17 (10)  
    RCMD 3 (13) 18 (11)  
    RCMD-RS 1 (4) 8 (5)  
    RAEB-1 1 (4) 21 (12)  
    RAEB-2 5 (21) 26 (15)  
    MDS-U 1 (4) 6 (4)  
    Missing data 3 (13) 24 (14)  
Karyotype risk, n (%)   .67 
    Good 16 (67) 93 (55)  
    Intermediate 2 (8) 18 (11)  
    Poor 2 (8) 21 (12)  
    Missing data 4 (17) 37 (22)  
BM blasts, n (%)   .35 
    < 5% 15 (62) 93 (55)  
    5%-9% 1 (4) 23 (14)  
    10%-19% 5 (21) 25 (15)  
    Missing data 3 (13) 28 (16)  
Hemoglobin, n (%)   .46 
    < 8 g/L 5 (21) 30 (18)  
    8-10 g/L 8 (33) 70 (41)  
    > 10 g/L 8 (33) 36 (21)  
    Missing data 3 (13) 33 (20)  
IPSS, n (%)   .37 
    Low-risk 4 (16) 35 (21)  
    Intermediate-1 10 (42) 47 (28)  
    Intermediate-2 3 (13) 35 (21)  
    High 3 (13) 10 (6)  
    Missing data 4 (16) 39 (24)  
Transfusion dependence, n (%)   .96 
    Yes 17 (71) 107 (63)  
    No 4 (16) 26 (16)  
    Missing data 3 (13) 36 (21)  
Transformation into AML, n (%)   .013 
    Yes 11 (46) 34 (20)  
    No 10 (41) 98 (58)  
    Missing data 3 (13) 37 (22)  
SF3B1 mutations, n (%)   .75 
    Mutated 4 (17) 24 (14)  
    WT 20 (83) 145 (86)  
    Missing data 0 (0) 0 (0)  
U2AF1 mutations, n (%)   .143 
    Mutated 0 (0) 14 (8)  
    WT 24 (100) 155 (92)  
    Missing data 0 (0) 0 (0)  
ZRSR2 mutations, n (%)   .35 
    Mutated 0 (0) 6 (4)  
    WT 24 (100) 163 (96)  
    Missing data 0 (0) 0 (0)  
ASXL1 fs mutations, n (%)   .58 
    Mutated 6 (25) 34 (20)  
    WT 18 (75) 135 (80)  
    Missing data 0 (0) 0 (0)  
RUNX1 mutations, n (%)   < .001 
    Mutated 7 (29) 7 (4)  
    WT 17 (71) 162 (96)  
    Missing data 0 (0) 0 (0)  
TP53 mutations, n (%)   .7 
    Mutated 0 (0) 4 (2)  
    WT 24 (100) 164 (97)  
    Missing data 0 (0) 1 (0)  
IDH1 mutations, n (%)   .013 
    Mutated 3 (13) 4 (2)  
    WT 21 (87) 165 (98)  
    Missing data 0 (0) 0 (0)  
DNMT3A mutations, n (%)   .39 
    Mutated 0 (0) 5 (3)  
    WT 24 (100) 164 (97)  
    Missing data 0 (0) 0 (0)  
NRAS mutations, n (%)   .31 
    Mutated 0 (0) 7 (4)  
    WT 24 (100) 162 (96)  
    Missing data 0 (0) 0 (0)  
Characteristic SRSF2 mutated (n = 24) SRSF2 WT (n = 169) P 
Age, y   .12 
    Median 69 65  
    Range 56-83 36-92  
Sex, n (%)   .059 
    Male 19 (79) 100  
    Female 5 (21) 69  
WHO subtype, n (%)   .86 
    RA 7 (29) 31 (18)  
    RARS 2 (8) 18 (11)  
    del (5q) 1 (4) 17 (10)  
    RCMD 3 (13) 18 (11)  
    RCMD-RS 1 (4) 8 (5)  
    RAEB-1 1 (4) 21 (12)  
    RAEB-2 5 (21) 26 (15)  
    MDS-U 1 (4) 6 (4)  
    Missing data 3 (13) 24 (14)  
Karyotype risk, n (%)   .67 
    Good 16 (67) 93 (55)  
    Intermediate 2 (8) 18 (11)  
    Poor 2 (8) 21 (12)  
    Missing data 4 (17) 37 (22)  
BM blasts, n (%)   .35 
    < 5% 15 (62) 93 (55)  
    5%-9% 1 (4) 23 (14)  
    10%-19% 5 (21) 25 (15)  
    Missing data 3 (13) 28 (16)  
Hemoglobin, n (%)   .46 
    < 8 g/L 5 (21) 30 (18)  
    8-10 g/L 8 (33) 70 (41)  
    > 10 g/L 8 (33) 36 (21)  
    Missing data 3 (13) 33 (20)  
IPSS, n (%)   .37 
    Low-risk 4 (16) 35 (21)  
    Intermediate-1 10 (42) 47 (28)  
    Intermediate-2 3 (13) 35 (21)  
    High 3 (13) 10 (6)  
    Missing data 4 (16) 39 (24)  
Transfusion dependence, n (%)   .96 
    Yes 17 (71) 107 (63)  
    No 4 (16) 26 (16)  
    Missing data 3 (13) 36 (21)  
Transformation into AML, n (%)   .013 
    Yes 11 (46) 34 (20)  
    No 10 (41) 98 (58)  
    Missing data 3 (13) 37 (22)  
SF3B1 mutations, n (%)   .75 
    Mutated 4 (17) 24 (14)  
    WT 20 (83) 145 (86)  
    Missing data 0 (0) 0 (0)  
U2AF1 mutations, n (%)   .143 
    Mutated 0 (0) 14 (8)  
    WT 24 (100) 155 (92)  
    Missing data 0 (0) 0 (0)  
ZRSR2 mutations, n (%)   .35 
    Mutated 0 (0) 6 (4)  
    WT 24 (100) 163 (96)  
    Missing data 0 (0) 0 (0)  
ASXL1 fs mutations, n (%)   .58 
    Mutated 6 (25) 34 (20)  
    WT 18 (75) 135 (80)  
    Missing data 0 (0) 0 (0)  
RUNX1 mutations, n (%)   < .001 
    Mutated 7 (29) 7 (4)  
    WT 17 (71) 162 (96)  
    Missing data 0 (0) 0 (0)  
TP53 mutations, n (%)   .7 
    Mutated 0 (0) 4 (2)  
    WT 24 (100) 164 (97)  
    Missing data 0 (0) 1 (0)  
IDH1 mutations, n (%)   .013 
    Mutated 3 (13) 4 (2)  
    WT 21 (87) 165 (98)  
    Missing data 0 (0) 0 (0)  
DNMT3A mutations, n (%)   .39 
    Mutated 0 (0) 5 (3)  
    WT 24 (100) 164 (97)  
    Missing data 0 (0) 0 (0)  
NRAS mutations, n (%)   .31 
    Mutated 0 (0) 7 (4)  
    WT 24 (100) 162 (96)  
    Missing data 0 (0) 0 (0)  

RA indicates refractory anemia; RARS, refractory anemia with ring sideroblasts; del(5q), MDS with isolated del(5q); RCMD, refractory cytopenia with multilineage dysplasia; RAEB-1, refractory anemia with excess blasts-1; RAEB-2, refractory anemia with excess blasts-2; MDS-U, MDS-unclassifiable; and fs, frameshift.

Cytogenetic and molecular analysis

Cytogenetic analysis was performed centrally by G- and R-banding analysis. Mutation analysis was performed as described previously.15  Mononuclear cells from patient samples were enriched by Ficoll density gradient centrifugation and were stored at −196°C in liquid nitrogen until use. Genomic DNA was extracted from samples using the All Prep DNA/RNA Kit (QIAGEN) according to the manufacturer's recommendations. Mutation analyses were performed for ASXL1,16 DNMT3A,17 IDH1,18 IDH2,19 RUNX1,20 NRAS,21 TP53,22 NPM1,21  and SF3B1,9  as described previously. PCR fragments were sequenced directly and analyzed using Sequencing Analysis Version 5.3.1 software (Applied Biosystems) and Vector NTI Advance 10 software (Invitrogen). All mutations were confirmed in an independent experiment.

Analysis of mutations in genes of the splicing machinery

Preparation of mononuclear cells and genomic DNA was performed as described previously.15,18  Genomic DNA was amplified using a whole-genome amplification kit following the manufacturer's recommendations (Sigma-Aldrich). All 11 exons of ZRSR2 and the mutational hotspots of U2AF1 (exon 2 and exon 6) and SRSF2 (exon 1) were sequenced in all MDS patients using primers as described in supplemental Table 1. Mutational hotspots of U2AF1 and SRSF2 were also sequenced in 50 healthy volunteers. Patient DNA was amplified using the following PCR conditions: 95°C for 10 minutes and 39 cycles of 95°C for 1 minute, 56°C for 1 minute, and 72°C for 70 seconds, followed by 1 cycle of 72°C for 10 minutes. Sequencing of SRSF2 required 2 rounds of amplification using nested primers with the same cycling conditions as above for 35 cycles with each primer pair (supplemental Table 1). Purified PCR fragments were sequenced directly using forward or reverse primers. The sequences were compared with the reference sequence available from Ensembl (transcripts ZRSR2 ENST00000307771, U2AF1 ENST00000291552, and SRSF2 ENST00000392485). All mutations were confirmed in nonamplified genomic DNA in an independent experiment. The somatic or germline status of mutations in U2AF1 and SRSF2 was established by evaluating T cells (CD3+CD11bCD14CD33) purified from diagnostic samples by flow cytometry.

Statistical analysis

Overall survival (OS) end points, measured from the date of first sample collection, were death (failure) and alive at last follow-up (censored). Time to AML progression was measured from date of first sample collection to the time of AML diagnosis. Progression to AML was defined according to the 2008 WHO classification. The median follow-up time for patients alive was calculated according to the method of Korn.23  Primary analysis was performed on OS and time to AML progression. The Kaplan-Meier method, log-rank test, and Cox proportional hazards models were used to estimate the distribution of OS and time to AML progression and to compare differences between survival curves, respectively. Pairwise comparisons were performed by Mann-Whitney test or the Student t test for continuous variables and by 2-sided χ2 tests for categorical variables and are provided for exploratory purposes. Associations between gene mutations are represented by a Circos diagram.24  For multivariate analysis, a Cox proportional hazards model was constructed for OS and time to progression to AML, adjusting for potential confounding covariates.25  Variables considered for model inclusion were IPSS, transfusion dependence, ferritin level (above or below 1000 μg/L), age (below vs above median), SF3B1 mutation status, ASXL1 mutation status, RUNX1 mutation status, IDH1 mutation status, TP53 mutation status, DNMT3A and NRAS mutation status, and SRSF2, U2AF1, and ZRSR2 mutation status. A limited backward-selection procedure was used to exclude redundant or unnecessary variables. Variables with P ≤ .1 in univariate analysis for OS were included in the model. The 2-sided level of significance was set at P < .05. The statistical analyses were performed with the statistical software package SPSS Version 19.0 (IBM Deutschland GmbH).

Results

Mutation status of SRSF2, U2AF1, ZRSR2, and SF3B1 in MDS patients

Mutations in one of the genes belonging to the splicing machinery (SRSF2, U2AF1, ZRSR2, and SF3B1) were found in 67 of 193 MDS patients (34.7%). Twenty-four patients were identified as harboring a mutation in SRSF2 (12.4%), 14 patients showed a mutation in U2AF1 (7.3%), and 6 patients were found to have a mutation in the coding region of ZRSR2 (3.1%). In addition, we identified intronic mutations in 2 patients at base pair 1 and 7, respectively, of intron 7/8 of ZRSR2. As described previously, we identified 28 mutations of SF3B1 in our cohort.9  No aberrations in the mutational hotspots of SRSF2 (exon 1), U2AF1 (exon 2 and exon 6), or SF3B1 (exons 13-16) were identified in any samples from healthy volunteers (n = 50). Interestingly, SRSF2, U2AF1, and ZRSR2 mutations were mutually exclusive, with no patient having more than one of these genes affected (Table 1 and supplemental Tables 2 and 3). However, we identified 4 patients with concomitant SF3B1 and SRSF2 mutations and 1 patient with concomitant SF3B1 and ZRSR2 mutations (Table 1 and supplemental Tables 2 and 3). The majority of mutations in SRSF2, U2AF1, and ZRSR2 were heterozygous point mutations (n = 33; Figure 1A); 9 frameshift and 2 nonsense mutations were also observed. In SRSF2, codon P95 located in exon 1 was most frequently mutated (n = 18), leading to a change from proline to leucine (n = 7), arginine (n = 6), or histidine (n = 5). Five patients harbored a frameshift mutation in codon P95 and 1 patient showed a heterozygous point mutation in codon P96 leading to a change from proline to leucine. In U2AF1, the most commonly affected codon was Q157 resulting in a glutamine to proline substitution in 9 patients. Four patients showed a heterozygous point mutation in codon S34 leading to a change from serine to phenylalanine and 1 patient had a frameshift mutation in codon E159. In ZRSR2, the only recurrent mutation identified was a frameshift in codon R448 (n = 3). However, 2 heterozygous nonsense mutations affecting codons R126 and R295 and 1 heterozygous point mutation causing a change from cysteine to serine in codon C181 were also identified. The somatic nature of mutations in U2AF1 and SRSF2 was confirmed by sequencing nontumoral CD3+ cells in 4 patients (2 patients with U2AF1 Q157P mutations and 2 patients with SRSF2 P95R mutations; supplemental Figures 1 and 2).

Figure 1

Gene mutations in patients with MDS. (A) Location and type of mutations in genes of the splicing machinery in 193 patients with MDS. (B) Associations of gene mutations in our MDS patient cohort outlined by a Circos diagram.

Figure 1

Gene mutations in patients with MDS. (A) Location and type of mutations in genes of the splicing machinery in 193 patients with MDS. (B) Associations of gene mutations in our MDS patient cohort outlined by a Circos diagram.

Patient characteristics in relation to SRSF2, U2AF1, and ZRSR2 mutation status

The clinical and hematologic characteristics of patients with mutated versus wild-type (WT) SRSF2, U2AF1, and ZRSR2 are shown in Table 1, supplemental Table 2, and supplemental Table 3, respectively. There were no significant differences in age, WHO classification, IPSS-based karyotype, BM blasts, hemoglobin, IPSS, or transfusion dependence between the mutated and the WT patients in any of the 3 genes. Interestingly, unlike SF3B1,9  we did not identify an association between SRSF2, U2AF1, or ZRSR2 mutation status and the presence of ring sideroblasts.

Patients with mutated SRSF2, but not U2AF1 or ZRSR2, were found to have a higher transformation rate to AML (P = .013). We found a correlation between mutations in genes belonging to the splicing machinery and other mutations (Figure 1B). For example, there was an association between SRSF2 mutations and RUNX1 mutations (P < .001) and IDH1 mutations (P = .013). Interestingly, U2AF1 mutations were associated with mutations in ASXL1 (P = .005) and DNMT3A (P = .004), both of which are involved in epigenetic regulation. Mutations of ZRSR2, which is located on the X chromosome, was found in both males (n = 4) and females (n = 2). We did not observe any association between ZRSR2 and other mutations, which may be because of the low number of ZRSR2-mutated patients in our cohort. Considering mutations in all genes investigated in our study, at least 1 mutation was detected in 100 (51.8%) patients, of which 20 (10.4%) had 2 mutated genes and 11 (5.7%) had 3 mutated genes.

Prognostic impact of SRSF2, U2AF1, and ZRSR2 mutations

The prognostic impact of mutations was evaluated in MDS patients for whom follow-up information was available (n = 154). The median follow-up of patients was 3 years. The prognostic impact of each mutated gene was calculated separately. Univariate analysis demonstrated that patients with SRSF2 mutations had significantly shorter OS compared with patients without the mutation (5-year OS, 17% vs 39%; hazard ratio [HR] = 1.76; 95% confidence interval [95% CI], 1.0-3.1; P = .049; Figure 2A and Table 2) and shorter time to AML progression (5-year AML-free survival 39% vs 69%; HR = 2.5; 95% CI, 1.22-5.1; P = .012; Figure 2B and Table 3). The relatively high incidence of mutations in SRSF2 in the IPSS low and intermediate-1 group (n = 14) allowed us to perform a subgroup analysis in patients with low and intermediate-1 IPSS. In this subgroup, SRSF2 mutations were associated with an inferior OS (5-year OS, 19% vs 50%; HR = 2.23; 95% CI, 1.11-4.5; P = .025) and shorter time to AML progression (5-year AML-free survival, 50% vs 81%; HR = 3.42; 95% CI, 1.14-10.27; P = .028; Figure 3A-B).

Figure 2

OS and time to AML progression in MDS patients. OS (A) and time to AML progression (B) in MDS patients according to SRSF2 mutation status (log-rank test) are shown. OS (C) and time to AML progression (D) in MDS patients according to U2AF1 mutation status (log-rank test) are shown.

Figure 2

OS and time to AML progression in MDS patients. OS (A) and time to AML progression (B) in MDS patients according to SRSF2 mutation status (log-rank test) are shown. OS (C) and time to AML progression (D) in MDS patients according to U2AF1 mutation status (log-rank test) are shown.

Table 2

Univariate and multivariate analysis for OS in MDS patients

 OS univariate analysis
 
OS multivariate analysis
 
HR* 95% CI P HR* 95% CI P 
SRSF2 mutation status, mutated vs WT 1.76 1.0-3.1 .049 2.3 1.28-4.13 .017 
ASXL1 mutation status, mutated vs WT 2.06 1.21-3.50 .008 2.01 1.13-3.57 .017 
RUNX1 mutation status, mutated vs WT 2.44 1.21-4.92 .013 1.28 0.60-2.71 .53 
IDH1 mutation status, mutated vs WT 3.28 1.48-7.05 .003 1.90 0.79-4.58 .16 
NRAS mutation status, mutated vs WT 5.39 2.29-12.66 < .001 5.25 1.99-13.84 .001 
Transfusion dependence, dependent vs independent 3.72 1.70-8.14 .001 2.79 1.23-6.36 .014 
IPSS, low vs intermediate 1 vs intermediate 2 vs high 1.73 1.35-2.2 < .001 1.48 1.13-1.94 .005 
 OS univariate analysis
 
OS multivariate analysis
 
HR* 95% CI P HR* 95% CI P 
SRSF2 mutation status, mutated vs WT 1.76 1.0-3.1 .049 2.3 1.28-4.13 .017 
ASXL1 mutation status, mutated vs WT 2.06 1.21-3.50 .008 2.01 1.13-3.57 .017 
RUNX1 mutation status, mutated vs WT 2.44 1.21-4.92 .013 1.28 0.60-2.71 .53 
IDH1 mutation status, mutated vs WT 3.28 1.48-7.05 .003 1.90 0.79-4.58 .16 
NRAS mutation status, mutated vs WT 5.39 2.29-12.66 < .001 5.25 1.99-13.84 .001 
Transfusion dependence, dependent vs independent 3.72 1.70-8.14 .001 2.79 1.23-6.36 .014 
IPSS, low vs intermediate 1 vs intermediate 2 vs high 1.73 1.35-2.2 < .001 1.48 1.13-1.94 .005 
*

HR > 1 indicates an increased risk of an event for the first category listed.

Table 3

Univariate and multivariate analysis for risk of AML progression in MDS patients

 AML transformation univariate analysis
 
AML transformation multivariate analysis
 
HR* 95% CI P HR* 95% CI P 
SRSF2 mutation status, mutated vs WT 2.5 1.22-5.1 .012 2.83 1.31-6.12 .008 
U2AF1 mutation status, mutated vs WT 2.53 0.90-7.13 .079 1.25 0.36-4.33 .73 
ASXL1 mutation status, mutated vs WT 2.3 1.15-4.62 .019 2.34 1.12-4.89 .024 
RUNX1 mutation status, mutated vs WT 3.02 1.26-7.26 .013 2.43 0.96-6.17 .062 
TP53 mutation status, mutated vs WT 1.003 1.001-1.005 .009 1.01 1.002-1.008 .001 
IDH1 mutation status, mutated vs WT 3.34 1.19-9.41 .023 2.33 0.75-7.21 .14 
NRAS mutation status, mutated vs WT 3.23 0.98-10.68 .054 1.4 .034-5.79 .64 
Transfusion dependence, dependent vs independent 6.77 1.63-28.09 .008 7.38 0.98-55.5 .052 
IPSS, low vs intermediate 1 vs intermediate 2 vs high 2.84 1.99-4.06 < .001 2.89 1.93-4.31 < .001 
 AML transformation univariate analysis
 
AML transformation multivariate analysis
 
HR* 95% CI P HR* 95% CI P 
SRSF2 mutation status, mutated vs WT 2.5 1.22-5.1 .012 2.83 1.31-6.12 .008 
U2AF1 mutation status, mutated vs WT 2.53 0.90-7.13 .079 1.25 0.36-4.33 .73 
ASXL1 mutation status, mutated vs WT 2.3 1.15-4.62 .019 2.34 1.12-4.89 .024 
RUNX1 mutation status, mutated vs WT 3.02 1.26-7.26 .013 2.43 0.96-6.17 .062 
TP53 mutation status, mutated vs WT 1.003 1.001-1.005 .009 1.01 1.002-1.008 .001 
IDH1 mutation status, mutated vs WT 3.34 1.19-9.41 .023 2.33 0.75-7.21 .14 
NRAS mutation status, mutated vs WT 3.23 0.98-10.68 .054 1.4 .034-5.79 .64 
Transfusion dependence, dependent vs independent 6.77 1.63-28.09 .008 7.38 0.98-55.5 .052 
IPSS, low vs intermediate 1 vs intermediate 2 vs high 2.84 1.99-4.06 < .001 2.89 1.93-4.31 < .001 
*

HR > 1 indicates an increased risk of an event for the first category listed.

Figure 3

OS and time to AML progression in MDS patients. OS (A) and time to AML progression (B) in MDS patients with IPSS low/intermediate-1 risk according to SRSF2 mutation status (log-rank test) are shown.

Figure 3

OS and time to AML progression in MDS patients. OS (A) and time to AML progression (B) in MDS patients with IPSS low/intermediate-1 risk according to SRSF2 mutation status (log-rank test) are shown.

Mutations in U2AF1 had no impact on OS (HR = 1.49; 95% CI, 0.54-4.1; P = .44; Figure 2C), but showed a trend toward a more rapid transformation to AML (HR = 2.53; 95% CI, 0.9-7.13; P = .079; Figure 2D and Table 3). However, this analysis was hampered by the small number of mutated patients (n = 9). The low incidence of ZRSR2 mutations precluded a formal survival analysis. Three mutated patients died after 56 days, 1.4 years, and 3.5 years, whereas 3 patients were alive at last follow up 117 days, 2.1 years, and 5.8 years after diagnosis. Multivariate analysis for OS, including mutations in SRSF2, ASXL1 (frameshift mutations), RUNX1, NRAS, and IDH1, as well as transfusion dependence and IPSS risk group, revealed that the presence of SRSF2 mutations was an independent unfavorable prognostic factor for OS (HR = 2.3; 95% CI, 1.28-4.13; P = .017; Table 2). Multivariate analysis for AML transformation, including mutations in SRSF2, U2AF1, ASXL1 (frameshift mutations), RUNX1, TP53, NRAS, and IDH1, as well as transfusion dependence and IPSS, demonstrated that the presence of SRSF2 but not U2AF1 mutations was found to be an independent unfavorable prognostic factor for AML transformation (SRSF2, HR = 2.83; 95% CI, 1.31-6.12; P = .008; Table 3; U2AF1, HR = 1.25; 95% CI, 0.36-4.33; P = .73; Table 3). The OS for patients with mutations in 1 of the splicing genes not associated with ring sideroblasts (ie, U2AF1, SRSF2, and ZRSR2) was significantly lower compared with WT patients (HR = 1.81; 95% CI, 1.11-2.98; P = .018; supplemental Figure 3A) and these patients had a significantly higher rate for AML transformation (HR = 2.33; 95% CI, 1.22-4.46; P = .011; supplemental Figure 3B). The OS and AML transformation rate was similar between mutated and WT patients when considering all 4 investigated splicing genes (SRSF2, U2AF1, ZRSR2, and SF3B1; OS, HR = 1.26; 95% CI, 0.8-1.99; P = .31; supplemental Figure 4A; AML transformation, HR = 1.28; 95% CI, 0.68-2.39; P = .48; supplemental Figure 4B).

Discussion

In this comprehensive analysis of 193 MDS patients, we identified mutations in at least 1 of the investigated genes involved in the splicing machinery (ie, SRSF2, U2AF1, ZRSR2, and SF3B1) in 34.7% of all investigated patients, suggesting that gene mutations involved in the splicing pathway are the most common mutations in MDS reported to date. Mutations in SF3B1 and SRSF2 were the most prevalent aberrations in MDS patients. U2AF1 mutations were slightly less frequent and ZRSR2 mutations were relatively rare. Unlike mutations in SF3B1,6,9  mutations in SRSF2, U2AF1, and ZRSR2 were not associated with the presence of ring sideroblasts. Instead, SRSF2, U2AF1, and ZRSR2 mutations occurred in all different subtypes of MDS and were not associated with a specific IPSS risk profile or cytogenetic aberration. An interesting finding in our analysis is that mutations in SRSF2 were associated with mutations in RUNX1, a gene coding for a transcription factor, as well as IDH1, a gene coding for an enzyme of the citric acid cycle. In contrast, mutations in U2AF1 were associated with ASXL1 and DNMT3A mutations, genes relevant for epigenetic regulation. The latter association is especially intriguing because chromatin and histone modifications have been shown to play a significant role in pre-mRNA splicing.26 

SRSF2 mutations were associated with a negative prognostic impact in our study. Patients with this mutation had a significantly inferior OS and a more rapid and more frequent progression to AML. This effect was also evident when considering patients with low-risk MDS (IPSS low and intermediate-1) alone, suggesting that SRSF2 mutations might be a molecular marker useful for risk stratification in the subgroup of lower-risk MDS patients. The negative prognostic impact was also seen in multivariate analysis both for OS and time to AML progression, further supporting the idea that SRSF2 mutations represent an independent prognostic marker in MDS. In accordance with an earlier study, we did not find an impact on OS for U2AF1 mutations.27  Similarly to Graubert et al, who reported that U2AF1 mutations were associated with an increased risk of progression to AML,27  we observed a trend toward a more rapid progression to AML for patients with U2AF1 mutations. However, the number of patients with U2AF1 mutations was small, so the result requires confirmation in a larger patient cohort. Interestingly, in our analysis, point mutations in codon S34 were less frequent than mutations in codon Q157, whereas Graubert et al found more mutations in codon S34.27 ZRSR2 mutations had no prognostic effect, possibly related to the low mutation rate. Some studies have reported that SF3B1 mutations are a favorable marker in MDS.6,8  However, other studies, including our own analysis, show that SF3B1 mutations do not represent an independent prognostic factor, but rather that the favorable prognosis of MDS patients with SF3B1 mutations may be a consequence of the favorable prognosis for MDS patients with ring sideroblasts.9,10  However, such a prognostic analysis of gene mutations in MDS is not only challenged by the heterogeneity of the disease, but potentially also by the different treatment strategies used. We evaluated OS in 1 uniformly treated patient group compared with all other patients; this comparison is hampered by several factors, including that patients were treated in different studies and that all lenalidomide-treated patients had 5q− syndrome, which has a better prognosis per se. Survival was similar in patients treated with antithymocyte globulin, thalidomide, or best supportive care compared with all other patients (data not shown). Lenalidomide-treated patients had an improved OS compared with all other patients (P = .027); however, all patients had 5q− syndrome, which has a better prognosis per se. Therefore, prospective, larger trials of uniformly treated patients are needed to confirm the effect of mutations on patient outcome.

The mechanisms of how different mutations in the same splicing pathway can have differing prognostic effects are unknown and need to be studied functionally. The presence of mutational hotspots in SRSF2, SF3B1, and U2AF1 suggests gain-of-function mutations in these genes, all of which encode for proteins belonging to the splicing E/A complex. U2AF1 and SRSF2 bind to the RNA directly, whereas ZRSR2 interacts with U2AF1 and SF3B1 binds to U2snRNP7 . In addition to participating in the splicing process, SRSF2 is also involved in the regulation of DNA stability.28  It was shown previously that depletion of SRSF2 contributes to genomic instability,28  which is a predictive marker for adverse outcome in MDS,29  and this could explain why SRSF2 mutations had a strong adverse impact. It was also shown previously that mutated U2AF1 has a growth-suppressive effect in vitro7  and this effect might contribute to the cytopenias seen in clinical MDS patients. The fact that in the present study, only SF3B1, not SRSF2, U2AF1, or ZRSR2, were associated with ring sideroblasts, suggests that the latter 3 mutations have a different pathophysiologic effect in the biology of MDS than SF3B1 mutations and underscores the idea that the genes in the splicing machinery have distinct biologic functions.

In summary, the results of the present study suggest that the splicing machinery is one of the most frequently affected pathways in MDS. Mutations in SRSF2 are independently associated with a negative prognosis for OS and AML transformation. The utility of SRSF2 mutations for treatment stratification should be assessed in future studies.

The online version of this article contains a data supplement.

The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Acknowledgments

The authors thank all patients and contributing doctors; Prof R. Blasczyk for the supply of healthy control samples; and the Cell Sorting Core Facility of the Hannover Medical School supported in part by Braukmann-Wittenberg-Herz-Stiftung and Deutsche Forschungsgemeinschaft.

This study was supported by the Deutsche Krebshilfe e.V (grant 109003); the Deutsche-José-Carreras Leukämie-Stiftung e.V (grant DJCLS R 10/22); H.W. & J. Hector Stiftung (grant M 47.1); Dieter-Schlag Stiftung (grant 2011); the German Federal Ministry of Education and Research (grant 01EO0802); the Deutsche Forschungsgemeinschaft (grants HE 5240/4-1 and TH 1779/1-1); and the Hannover Medical School (HiLF grant to F.T.)

Authorship

Contribution: F.T., A.G., and M.H. designed the research; F.T., S.K., C.S., P.L., M.M., M.W.W., B.K., M.W., K.G., and G.G. performed the research; F.T., J.K., G.B., O.O., W.-K.H., A.G., and M.H. contributed patient samples and clinical data; G.G. and B.S. performed cytogenetic studies; F.T., S.K., C.S., P.L., M.M., M.W.W., B.K., M.W., K.G., G.G., G.B., O.O., C.M.N., W.-K.H., B.S., A.G. and M.H. analyzed the data; F.T., A.G., and M.H. wrote the manuscript; and all authors read and agreed to the final version of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Dr Felicitas Thol, Department of Hematology, Hemostasis, Oncology, and Stem Cell Transplantation, Hannover Medical School, Carl-Neuberg Str 1, 30625 Hannover, Germany; e-mail: thol.felicitas@mh-hannover.de.

References

References
1
Jädersten
M
Hellstrom-Lindberg
E
Myelodysplastic syndromes: biology and treatment.
J Intern Med
2009
, vol. 
265
 
3
(pg. 
307
-
328
)
2
Bejar
R
Levine
R
Ebert
BL
Unraveling the molecular pathophysiology of myelodysplastic syndromes.
J Clin Oncol
2011
, vol. 
29
 
5
(pg. 
504
-
515
)
3
Bejar
R
Stevenson
K
Abdel-Wahab
O
, et al. 
Clinical effect of point mutations in myelodysplastic syndromes.
N Engl J Med
2011
, vol. 
364
 
26
(pg. 
2496
-
2506
)
4
Thol
F
Winschel
C
Ludeking
A
, et al. 
Rare occurrence of DNMT3A mutations in myelodysplastic syndromes.
Haematologica
2011
, vol. 
96
 
12
(pg. 
1870
-
1873
)
5
Marcucci
G
Haferlach
T
Dohner
H
Molecular genetics of adult acute myeloid leukemia: prognostic and therapeutic implications.
J Clin Oncol
2011
, vol. 
29
 
5
(pg. 
475
-
486
)
6
Papaemmanuil
E
Cazzola
M
Boultwood
J
, et al. 
Somatic SF3B1 mutation in myelodysplasia with ring sideroblasts.
N Engl J Med
2011
, vol. 
365
 
15
(pg. 
1384
-
1395
)
7
Yoshida
K
Sanada
M
Shiraishi
Y
, et al. 
Frequent pathway mutations of splicing machinery in myelodysplasia.
Nature
2011
, vol. 
478
 
7367
(pg. 
64
-
69
)
8
Malcovati
L
Papaemmanuil
E
Bowen
DT
, et al. 
Clinical significance of SF3B1 mutations in myelodysplastic syndromes and myelodysplastic/myeloproliferative neoplasms.
Blood
2011
, vol. 
118
 
24
(pg. 
6239
-
6246
)
9
Damm
F
Thol
F
Kosmider
O
, et al. 
SF3B1 mutations in myelodysplastic syndromes: clinical associations and prognostic implications [published online ahead of print November 8, 2011].
Leukemia
 
10
Patnaik
MM
Lasho
TL
Hodnefield
JM
, et al. 
SF3B1 mutations are prevalent in myelodysplastic syndromes with ring sideroblasts but do not hold independent prognostic value.
Blood
2012
, vol. 
119
 
2
(pg. 
569
-
572
)
11
Hofmann
WK
Ganser
A
Seipelt
G
, et al. 
Treatment of patients with low-risk myelodysplastic syndromes using a combination of all-trans retinoic acid, interferon alpha, and granulocyte colony-stimulating factor.
Ann Hematol
1999
, vol. 
78
 
3
(pg. 
125
-
130
)
12
Stadler
M
Germing
U
Kliche
KO
, et al. 
A prospective, randomised, phase II study of horse antithymocyte globulin vs rabbit antithymocyte globulin as immune-modulating therapy in patients with low-risk myelodysplastic syndromes.
Leukemia
2004
, vol. 
18
 
3
(pg. 
460
-
465
)
13
Passweg
JR
Giagounidis
AA
Simcock
M
, et al. 
Immunosuppressive therapy for patients with myelodysplastic syndrome: a prospective randomized multicenter phase III trial comparing antithymocyte globulin plus cyclosporine with best supportive care–SAKK 33/99.
J Clin Oncol
2011
, vol. 
29
 
3
(pg. 
303
-
309
)
14
Porter
J
Oral iron chelators: prospects for future development.
Eur J Haematol
1989
, vol. 
43
 
4
(pg. 
271
-
285
)
15
Damm
F
Heuser
M
Morgan
M
, et al. 
Single nucleotide polymorphism in the mutational hotspot of WT1 predicts a favorable outcome in patients with cytogenetically normal acute myeloid leukemia.
J Clin Oncol
2010
, vol. 
28
 
4
(pg. 
578
-
585
)
16
Thol
F
Friesen
I
Damm
F
, et al. 
Prognostic significance of ASXL1 mutations in patients with myelodysplastic syndromes.
J Clin Oncol
2011
, vol. 
29
 
18
(pg. 
2499
-
2506
)
17
Thol
F
Damm
F
Lüdeking
A
, et al. 
Incidence and prognostic influence of DNMT3A mutations in acute myeloid leukemia.
J Clin Oncol
2011
, vol. 
29
 
21
(pg. 
2889
-
2896
)
18
Wagner
K
Damm
F
Gohring
G
, et al. 
Impact of IDH1 R132 mutations and an IDH1 single nucleotide polymorphism in cytogenetically normal acute myeloid leukemia: SNP rs11554137 is an adverse prognostic factor.
J Clin Oncol
2010
, vol. 
28
 
14
(pg. 
2356
-
2364
)
19
Thol
F
Damm
F
Wagner
K
, et al. 
Prognostic impact of IDH2 mutations in cytogenetically normal acute myeloid leukemia.
Blood
2010
, vol. 
116
 
4
(pg. 
614
-
616
)
20
Gaidzik
VI
Bullinger
L
Schlenk
RF
, et al. 
RUNX1 mutations in acute myeloid leukemia: results from a comprehensive genetic and clinical analysis from the AML study group.
J Clin Oncol
2011
, vol. 
29
 
10
(pg. 
1364
-
1372
)
21
Schlenk
RF
Dohner
K
Krauter
J
, et al. 
Mutations and treatment outcome in cytogenetically normal acute myeloid leukemia.
N Engl J Med
2008
, vol. 
358
 
18
(pg. 
1909
-
1918
)
22
Kaneko
H
Misawa
S
Horiike
S
Nakai
H
Kashima
K
TP53 mutations emerge at early phase of myelodysplastic syndrome and are associated with complex chromosomal abnormalities.
Blood
1995
, vol. 
85
 
8
(pg. 
2189
-
93
)
23
Korn
EL
Censoring distributions as a measure of follow-up in survival analysis.
Stat Med
1986
, vol. 
5
 (pg. 
255
-
260
)
24
Krzywinski
M
Schein
J
Birol
I
, et al. 
Circos: an information aesthetic for comparative genomics.
Genome Res
2009
, vol. 
19
 
9
(pg. 
1639
-
1645
)
25
COX
D
Regression models and life tables.
J R Stat Soc B
1972
, vol. 
34
 (pg. 
187
-
202
)
26
Hnilicová
J
Stanek
D
Where splicing joins chromatin.
Nucleus
2011
, vol. 
2
 (pg. 
182
-
188
)
27
Graubert
TA
Shen
D
Ding
L
, et al. 
Recurrent mutations in the U2AF1 splicing factor in myelodysplastic syndromes.
Nat Genet
2012
, vol. 
44
 
1
(pg. 
53
-
57
)
28
Xiao
R
Sun
Y
Ding
JH
, et al. 
Splicing regulator SC35 is essential for genomic stability and cell proliferation during mammalian organogenesis.
Mol Cell Biol
2007
, vol. 
27
 
15
(pg. 
5393
-
5402
)
29
Heilig
CE
Loffler
H
Mahlknecht
U
, et al. 
Chromosomal instability correlates with poor outcome in patients with myelodysplastic syndromes irrespectively of the cytogenetic risk group.
J Cell Mol Med
2010
, vol. 
14
 
4
(pg. 
895
-
902
)

Author notes

*

A.G. and M.H. are co–senior authors.