Clinical significance of CSF3R, SRSF2 and SETBP1 mutations in chronic neutrophilic leukemia and chronic myelomonocytic leukemia

Chronic neutrophilic leukemia (CNL) and chronic myelomonocytic leukemia (CMML) are rare hematologic neoplasms. We performed CSF3R, SRSF2 and SETBP1 mutational analyses in 10 CNL and 56 CMML patients. In this sample cohort, 80% of CNL patients harbored CSF3R mutations, of which the CSF3R T618I mutation was dominant. Mutations in CSF3R and SETBP1 were found in 7.1% and 5.3% CMML patients respectively, while 25% of CMML patients carried SRSF2 mutations. Strikingly, we identified that all of the CSF3R mutations detected in CMML patients were represented by a P733T mutation. The CSF3R P733T mutation represents a novel CSF3R mutation. In addition, none of the four CSF3R P733T mutated patients carried SRSF2 mutations [0/14 (0%) patients with combined CSF3R P733T and SRSF2 mutations vs. 4/42 (9.5%) with CSF3R P733T and wt SRSF2, P < 0.001]. Both mut SRSF2 and mut SETBP1 patients had shorter overall survival (OS) and progression-free survival (PFS) compared to patients with wt SRSF2 (P < 0.001 both) and wt SETBP1 (P < 0.001 and P = 0.02, respectively). While we found no significant differences in OS and PFS as a consequence of CSF3R mutation status, our work suggest that the CSF3R T618I mutation is a diagnostic marker with good specificity and sensitivity for CNL. In conclusion, our study highlights effective diagnostic and prognostic markers of CNL and CMML patients in the Chinese population.


INTRODUCTION
CNL is a rare hematologic neoplasm that is diagnosed largely based on exclusion of underlying causes of reactive neutrophilia and/or the lack of specific molecular markers of other hematological malignancies [1]. Until the recent discoveries of CSF3R and SETBP1 mutations [2], no recurrent genetic abnormalities have been identified in CNL.
CSF3R (G-CSF-R), the colony-stimulating factor 3 receptor, is a trans-membrane protein which plays a prominent role in the growth and differentiation of granulocytes [3]. While CSF3R mutations are most commonly found in severe congenital neutropenia (SCN), the rates of CSF3R mutations rises sharply upon progression to secondary acute myeloid leukemia (sAML) [4][5][6]. Thus, CSF3R mutations may critically influence on disease progression to AML, although the types of truncation mutations that associate with sAML are rarely detected in other disorders, including de novo AML6. Recently, acquired CSF3R mutations (in particular the CSF3R T618I mutation) were described in a majority of patients diagnosed with CNL or atypical chronic myeloid leukemia (aCML) [2,7]. While the relevance of CSF3R mutations in CMML has also been investigated, discrepancies exist among different studies. For example, Kosmider et al. identified about 3% of CMML patients with CSF3R somatic mutations, while Pardanani et al failed to identify CSF3R mutations in CMML [8,9].
Here, we have unveiled the frequency, clinical significance and prognostic relevance of mutations in CSF3R, SETBP1 and SRSF2 in a cohort of 10 CNL and 56 CMML patients, with the aim of providing insights into the development of effective diagnostic and prognostic tools of CNL and CMML patients in the Chinese population.
Patients with CSF3R mutation had higher HB levels (P = 0.05) compared to wt CSF3R patients. However, no statistically significant difference was identified between mut CSF3R and wt CSF3R CMML patients in terms of age, gender, WHO category, FAB category, karyotype, other blood cell counts and CPSS risk stratification. In terms of treatment for the investigated patients, we normally used hydroxyurea and Interferon-α for CNL patients, and demethylation therapy, combined chemotherapy, hydroxyurea, immunomodulation therapy and supportive treatment for CMML patients.

SRSF2 and SETBP1 mutations are independent predictors of poor survival for CMML patients
We next focused our prognostic analysis on the impact of mutation status in CMML patients. Overall, during a median follow-up period of 20 months (range: 2-60 months), 14(25%) patients died, mainly due to progression to AML(10/14). Both SRSF2 and SETBP1 mutated patients showed shorter progression-free survival (PFS) and overall survival (OS) compared with wt SRSF2 (both P < 0.001) and wt SETBP1 (P = 0.02 and P < 0.001) patients. In addition, we observed that patients with CSF3R mutations had longer PFS and OS compared to wt CSF3R patients. However, the difference was not statistically significant (P > 0.05) ( Figure 3).

DISCUSSION
Next generation sequencing studies have revealed a large number of mutations in genes such as TET2, CBL, AXSL1, RUNX1, EZH2, RAS, JAK2, IDH1/IDH2, NPM1 and spliceosome mutations in CMML [18] as well as CSF3R, SETBP1 mutations in CNL and aCML [19]. In this study, we investigated the frequencies, diagnostic significance and clinical outcome of CSF3R, SETBP1 and SRSF2 gene mutations in CNL and CMML patients.
Our study show that 80% of CNL patients carried CSF3R mutations, with the substitution mutation of CSF3R T618I occurring exclusively in WHO-defined CNL.  Oncotarget 20839 www.impactjournals.com/oncotarget which did not affect survival [9]. In contrast, patients with SETBP1 mutations (33% in CNL) experienced shortened survival [9]. Because of the limited number of mutated cases, we only described the clinical characteristics of CNL patients without analyzing the prognostic status of these patients (Table 2).
We identified CSF3R mutations in 7.1% of CMML patients. Interestingly, all of the CSF3R mutations detected in CMML patients were CSF3R P733T mutations. CSF3R P733T mutation represents a novel CSF3R point mutation not previously reported. In addition, none of the 4 CSF3R P733T mutated patients were detected with SRSF2 mutations, suggesting that CSF3R P733T mutations and SRSF2 mutations are mutually exclusive in CMML patients. Therefore, the CSF3R P733T mutation may be of diagnostic value in CMML, especially for wt SRSF2 CMML patients. In univariate analysis, we observed that CMML patients with CSF3R mutations had longer PFS and OS compared to wt CSF3R patients. However, the difference was not statistically significant. The lack of significance may be due to the small sample size in the CSF3R mutant group which resulted in low statistical power. In contrast, Kosmider and collegues identified CSF3R mutations in 3% CMML patients and showed that CSF3R mutations were associated with an reduced OS and AML-free survival in univariate analysis [8]. Additionally, Maxson et al. showed that CSF3R mutations were divided into membrane proximal mutations and truncation mutations based on the distribution within two distinct regions of CSF3R [2]. Truncation mutations in CSF3R activates downstream signaling mediators-SRC family kinases (SFKs) and TNK2 and lead to increased sensitivity to Dasatinib treatment [2]. Membrane proximal mutations, on the other hand, showed preferential activation of JAK signaling pathway and susceptible to JAK kinase inhibitors such as Ruxolitinib [2]. Fundamental differences in mutations types might therefore indicate differences in prognosis of CMML patients, which deserves further attention.
Here, we demonstrate SETBP1 mutations present in 5.3% of CMML patients. In univariate analysis, SETBP1 mutations was an independent adverse predicting factor for OS (P = 0.002) and PFS (P = 0.038). Similar results were obtained for OS in multivariate analysis. These data are consistent with multiple previous studies [15,16,[19][20][21], which have shown that SETBP1 mutations occur in 4-7% CMML patients and are indicative of decreased OS and AML-free survival [22].
Notably, SRSF2 mutations were detected in 25% CMML patients in our study but exclusive of mutations in CSF3R. Our multivariate analysis showed SRSF2 mutation as an independent poor predictor for OS (P = 0.028) and PFS (P = 0.001), in line with previous data from Itzykson et al. and Makishima et al. [11,23].
In conclusion, we report that the majority of CNL patients studied here carried oncogenic CSF3R mutations and that the CSF3R T618I mutation appears to be a diagnostic marker with good specificity and sensitivity for CNL. By contrast, CSF3R P733T mutations detected in CMML patients were completely different from CSF3R mutation types described in patients with CNL, SCN and hereditary neutrophilia, which may be a potential diagnostic marker, particularly for wt SRSF2 CMML patients. In addition, mutations in SRSF2 were commonly found in CMML patients, and represents a poor prognostic marker for CMML. Mutations in SETBP1, by contrast, were found in CNL, CMML but also other hematological malignancies, making it a rather poor isolated prognostic marker for hematological diseases. As patients with different gene mutations may have different clinical response to treatment, mutation-based personalized targeted therapy should be considered in future studies.

Study population and definitions
A total of 10 CNL patients and 56 CMML patients were included in the study. 43 were male and 23 were female, at a median age of 64 (range: 24-94) years. Of the 56 CMML patients, 47 were diagnosed with CMML-1 and 9 were diagnosed with CMML-2. Of the 10 CNL patients, 1 was diagnosed with monoclonal gammopathy of undetermined significance associated with chronic neutrophilic leukemia (MGUS-CNL). 20 MDS patients, 10 chronic eosinophilic leukemia (CEL) patients and 20 healthy donors were included as controls. The diagnosis, classification and AML transformation of CMML, CNL, MDS and CEL were based on WHO 2008 criteria [1].

CSF3R, SRSF2 and SETBP1 mutation analysis
We obtained DNA specimens from bone marrow mononuclear cells or peripheral-blood granulocytes from the patients, and constitutional DNA samples from matched buccal swabs as described previously [9]. CSF3R exons 14-17, SETBP1 exon 4 and SRSF2 exon 1 were amplified. Independent validations of the detected variants were conducted using Sanger sequencing. If nonsynonymous sequence changes were detected, we www.impactjournals.com/oncotarget determined whether the sequence variant had previously been reported as a SNP by searching in the Single Nucleotide Polymorphism database (https://www.ncbi. nlm.nih.gov/projects/SNP/). The previously unreported variants were resequenced with the matched buccal swabs samples DNA from the same patients. The somatic mutation was confirmed when it associated with the bone marrow or peripheral-blood sample DNA and not the matched buccal swabs DNA.

Study of clinical features
Patient data was collected at the first diagnosis. The median follow-up period was 20 months (range: 2-60). CMML patients were divided into high, intermediate-1, intermediate-2 and low risk group according to the CMML Prognostic Scoring System (CPSS) [24].

Statistical analysis
The statistical analysis of data was done by using Excel and SPSS (statistical package for social science) version 17.0. Statistical analysis was performed by comparison between groups using Kruskal-Wallis test regarding quantitative nonparametric data and chi-square test regarding qualitative data. The overall survival analysis was done by Kaplan-Meier curve. Multivariate analysis of survival was carried out by Cox regression. All P-values < 0.05 (two-tailed) were considered statistically significant.