SRSF2 and U2AF1 mutations in primary myelofibrosis are associated with JAK2 and MPL but not calreticulin mutation and may independently reoccur after allogeneic stem cell transplantation.

Spliceosome gene mutations are among the 50–60 driver mutations underlying myelodysplastic syndromes (MDSs).[1] U2AF1 mutations for example have been reported to occur in up to 16% of primary myelofibrosis (PMF), and was found to be associated with anemia and thrombocytopenia in PMF.[2] We could show that spliceosome gene mutations are already present in early stages of PMF before fibrosis and cytopenia become manifest.[3] Recently, a negative association between mutations of calreticulin (CALR) and spliceosome genes has been described.[4]

CALR is a Ca2+-binding protein, which was found in 2013 to be mutated in JAK2- or MPL-unmutated PMF and essential thrombocythemia.[5,6] Mutations were mutually exclusive of JAK2 or MPL mutations. JAK2-mutated and triple-negative patients were shown to have significantly shorter survival periods in comparison to those with somatic frameshift mutations in the CALR gene.[4,5] Tefferi et al.[4] described significantly lower frequency of spliceosome mutations in CALR-mutated cases and attributed the lower incidence of anemia to the lower frequency of U2AF1 mutations.

Up to now allogeneic hematopoietic stem cell transplantation (AHSCT) represents the only curative treatment modus for patients with PMF.[7] Selection of patients suitable for this kind of treatment is performed according to prognostic scoring and tolerable risks of individual patients. Data of Heuser et al.[8] suggest a better overall survival for CALR-mutated PMF patients after AHSCT.

In this study, we analyzed 69 patients with PMF grades of fibrosis 2–3 (Table 1) who have undergone allogenic stem cell transplantation for JAK2, MPL, CALR and spliceosome gene mutations (SRSF2, U2AF1 and SF3B1) using bone marrow trephines and pyrosequencing as described.[9]

Table 1

PMF patients treated by allogeneic stem cell transplantation

	Patients with AHSCT	Patients with splice factor gene mutations	Patients with follow-up biopsies (n=52) and molecular relapse
Number of patients (%)	69	23 (33%)	4 (7.7%)
Median age, years (range)	65.5 (33–75)	66 (44–76)	64 (61–76)
Male	42	15	3
Female	27	8	1
JAK2V617F (%)	41 (59.4%)	18 (78.2%)	3
		5 JAK2/SRSF2
		10 JAK2/U2AF1
		3 JAK2/SF3B1
MPL (exon 10)	2 (2.9%)	1 (1.4%)	0
		1 MPL/SF3B1
Calreticulin (exon 9) (%)	19 (27.5%)	2 (8.7%)	0
		2 CALR/SF3B1
SRSF2 (exon 1) (%)	7 (10.1%)	7 (21.8%)	2
		5 SRSF2/JAK2
		2 SRSF2/TN
U2AF1 (exons 2, 6) (%)	9 (13.0%)	10 (43.5%)	1
		10 U2AF1/JAK2
SF3B1 (exons 14, 15)	6 (8.7%)	6 (26.1%)	0
		3 SF3B1/JAK2
		1 SF3B1/MPL
		2 SF3B1/CALR
Median time of follow-up biopsy (months)	13	14	18
Myelofibrosis grade 2 and 3 (%)	69 (100%)	23 (100%)	2 (50%)

Abbreviation: AHSCT, allogeneic hematopoietic stem cell transplantation; PMF, primary myelofibrosis; TN, triple negative.

CALR was rarely combined with splice factor gene mutations (10.5% of all CALR-mutated cases; negative correlation, P=0.0418) and these combinations were restricted to SF3B1. Combined mutations with U2AF1 and SRSF2 could not be found at all. The frequency of accompanying splice factor gene mutations in CALR-mutated patients was significantly lower than that in patients without a CALR mutation (21/50, 42% P=0.04) or in those with a JAK2 mutation (18/41, 44% P=0.04). U2AF1 was the most frequent splice factor gene mutation associated with JAKV617F. In PMF, splice factor gene mutations were associated significantly more often with a JAK2 mutation than with a CALR mutation (P<0.00005; Fisher's exact tests).

In our cohort 7 patients (10%) revealed neither JAK2 nor MPL or CALR mutation. In the ‘triple-negative' subgroup of PMF, exclusively mutations of SRSF2 occurred (n=2), but no alterations of U2AF1 and SF3B1 could be observed (Table 1).

Because of the low number of MPL-mutated cases in this series additional samples of PMF with bone marrow fibrosis grade 2–3 and known MPL mutation (n=20, all JAK2 exon 14 wild type) were investigated for combination with splice factor gene mutations. Among the 20 MPL-mutated cases 10 samples exhibited splice factor gene mutations (50%). Three samples revealed mutation of U2AF1 (15%), six of SRSF2 (30%) and one of SF3B1 (5%), respectively. Consequently, MPL-mutated PMF cases appear to carry splice factor gene mutations with a similar frequency as JAK2-mutated cases. CALR-mutated cases behave different from JAK2- and MPL-mutated cases in that splice factor gene mutations occur significantly rarer (P<0.005 for MPL) and only combinations with SF3B1 could be found.

After a median follow-up of 18 months four patients suffered molecular and histopathological relapse. Interestingly, the recurrent disease was different from the primary MPN and differences with regard to histopathology as well as to molecular aberrations could be observed. In two patients the bone marrow displayed reduced myelofibrosis (cytopenia was still evident (Patient 3 and 4 in Table 2). JAK2 mutation was no longer detectable in these cases but splice factor gene mutations persisted and as shown for SRSF2 to a similar allelic burden as in the bone marrow before AHSCT. Bone marrow biopsy in one case (patient 3) still revealed atypical megakaryocytes and myelofibrosis (MF1), consistent with relapse of PMF but with reduction of fibrosis (Table 2). In another case with persisting U2AF1 mutation after AHSCT histology showed a different picture more reminiscent of MDS with excess of blasts, and megakaryocytes did not show the atypia anymore characteristic for PMF (patient 4, Table 2). In patient No 1 JAK2 mutation was still detectable, but with a reduced allelic burden (15%) compared with the primary biopsy (80%), whereas SRSF2 mutation remained on an identical level of 50% (Table 2). In this case myelofibrosis was also diminished 1 year after AHSCT (MF grade 3 to MF grade 1) but increased again to MF grade 2 after 4 years. In addition, there was one case with persisting JAK2 mutation and loss of splice factor gene mutations (U2AF1) after AHSCT (patient 2).

Table 2

Molecular relapses after allogeneic stem cell transplantation of PMF

	P1	P1+AHSCT	P2	P2+AHSCT	P3	P3+AHSCT	P4	P4+AHSCT
JAK2	V617F	V617F	V617F	V617F	V617F	WT	V617F	WT
	80%	15%	50%	14%	14%		27%
SRSF2	P95H	P95H	WT	WT	P95H	P95H	WT	WT
	50%	50%			30%	40%
U2AF1	WT	WT	Q157P 40%	WT	WT	WT	Q157P 40%	Q157P 40%
Myelofibrosis	MF3	MF2	MF3	MF3	MF3	MF1	MF3	MF0
Hemoglobin (g/dl)	9.2	11.5	7.1	8.4	9.4	10.1	8.4	8.6
Leukocytes ( × 103/μl)	13.7	13.8	3.8	0.2	4.4	3.9	21.6	1.7
Thrombocytes ( × 103/μl)	42	112	72	12	115	21	32	9

Abbreviations: +AHCST, after allogeneic hematopoietic stem cell transplantation; MF 0–3, grade of myelofibrosis; P, patient; PMF, primary myelofibrosis; WT, wild type.

Our results show that PMF with high-risk scores eligible for AHSCT represents a molecularly heterogeneous disease despite uniform histopathology of bone marrow with atypical megakaryocytic proliferation and evident myelofibrosis. Cytopenia which is used for risk stratification and which is the dominant cause to treat patients with AHSCT appears to be associated in a considerable proportion of PMF cases with splice factor gene mutations. In these cases fibrotic obliteration of bone marrow spaces seems not to be the only cause of cytopenia. Splice factor gene mutations are significantly more frequently combined with JAK2 and MPL mutations than with CALR mutation. Furthermore, different hematopoietic clones proliferate in PMF giving rise to a molecular mosaic. After AHSCT of PMF relapses may uncover the underlying clonal mosaic and different diseases may emerge. Despite reduction of myelofibrosis and eradication of the JAK2V617F clone cytopenia may persist. The molecular mosaic in myeloproliferative neoplasms has been demonstrated to be the result of both independent clones proliferating in parallel as well as clonal evolution with stepwise acquisition of different mutations by a single neoplastic clone.[10] Molecular monitoring of patients having undergone AHSCT for PMF should not be restricted to JAK2, MPL or CALR, but all mutations present in the primary fibrotic neoplastic myeloproliferation should be included to interpret abnormal blood values after AHSCT. The apparently better prognosis of CALR-mutated PMF.[4,5] including cases treated with AHSCT[8] may at least in part be attributable to a less likely association with splice factor gene mutations.