Abstract
Mutations in genes encoding RNA splicing factors have been recently identified in myelodysplastic syndromes (MDS) and myelodysplastic/myeloproliferative neoplasms (MDS/MPN). SRSF2 mutations have been reported in up to 20% of patients with MDS and in about 40 to 50% of those with chronic myelomonocytic leukemia (CMML). In this latter condition, SRFS2 is the second recurrently mutated gene after TET2, and a significant association between TET2 and SRSF2 mutations has been reported. In MDS, SRSF2 mutations have been associated with a poor prognosis and shortened overall survival and leukemia-free survival, but no clear correlation has been described so far with specific disease phenotype.
In this work, we studied a large and well characterized cohort of MDS and MDS/MPN patients with the aim to describe clinical and molecular features of myeloid neoplasms carrying SRSF2 mutations, to recognize patterns of co-occurring or mutually exclusive mutations, and to identify correlations with disease phenotypes.
Massively parallel pyrosequencing was used to screen coding exons of 111 genes implicated in myeloid malignancies in 267 patients with MDS or MDS/MPN according to WHO criteria, diagnosed at the Department of Hematology, Fondazione IRCCS Policlinico S. Matteo, University of Pavia, Italy. Two-hundred forty five patients were classified as MDS (26 RA; 35 RARS; 69 RCMD; 27 RCMD-RS; 10 MDS with isolated del(5q); 36 RAEB-1 and 42 RAEB-2), and 22 as MDS/MPN (20 CMML and 2 MDS/MPN, unclassifiable, MDS/MPN-U).
High-confidence oncogenic mutations were identified in 43 genes, with a median number of 2 mutations per patient (range 0-9).
In the MDS/MPN cohort, the most frequent mutations involved SRSF2 (10/22, 45%), TET2 (10/22, 45%) and ASXL1 (5/22, 23%). Seven of 10 (70%) patients with SRSF2 mutation also had a mutation in TET2. SRSF2-mutated patients showed significantly higher hemoglobin level (12.1 vs 9.3 g/dL, P<.001 ) and higher monocyte count (2.42 vs 2.18 x109/L, P=.007) compared to those wild-type for SRSF2. Only 1 of 10 (10%) patients with SRSF2 mutations showed ring sideroblasts in the bone marrow.
Among MDS patients, mutations of SRSF2 were observed in 34/245 (14%); significant associations were found with ASXL1 ( 9/34, 26%, P=.027) and IDH 1/2 (9/34, 26%, P<.001). Conversely, mutations of DNMT3A and SF3B1 were mutually exclusive with SRSF2 mutations (0/34, P<.001), while no significant association with TET2 mutations was demonstrated (6/34, 18%, P=ns). A significant association was found between SRSF2 mutations and multilineage dysplasia or excess blasts (2/34 patients with unilineage dysplasia, 16/34 with multilineage dysplasia and 16/34 with RAEB, P=.028). Moreover, SRSF2 mutations were significantly associated with the presence of a variable proportion of bone marrow ring sideroblasts (range 1-68%) (17 of 34 patients, 50%, P=.04). The cases with ring sideroblasts were classified according to WHO criteria as RA or RARS (2), RCMD or RCMD-RS (9), RAEB-1 or RAEB-2 (6). No significantly different mutation pattern was observed in SRSF2-mutated patients with or without ring sideroblasts.
Finally, we investigated with an unsupervised approach in our whole cohort of myeloid neoplasms whether mutations of SRSF2 isolated or in association with co-occurring mutations could generate specific disease phenotypes. We found that the association of SRSF2 and TET2 mutations was highly specific for CMML disease phenotype (specificity 97.5%). Interestingly, of the six double mutated patients with a diagnosis of MDS, two showed relative monocytosis at the time of the analysis and two developed an overt CMML during follow-up. When accounting for these cases, the co-occurrence of SRSF2 and TET2 mutations showed specificity and positive predictive value for myelomonocytic phenotype of 99.2% and 84.6% respectively.
SRSF2 mutations are significantly associated with multilineage dysplasia and bone marrow ring sideroblasts in MDS patients, whereas the co-occurrence of SRSF2 and TET2 mutations is highly specific for the myelomonocytic phenotype of CMML. These data suggest that SRSF2 mutation is associated per se with a specific phenotype (ring sideroblasts and multilineage dysplasia), but also that its phenotypic impact may be modulated through interaction with other mutant genes like TET2.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.