High frequencies of SF3B1 and JAK2 mutations in refractory anemia with ring sideroblasts associated with marked thrombocytosis strengthen the assignment to the category of myelodysplastic/myeloproliferative neoplasms

Mutations of spliceosome genes were shown to occur frequently in different entities.1 Remarkably, mutations in SF3B1 (splicing factor 3b, subunit 1) were associated with the morphological feature of ring sideroblasts1-3 and were also found in refractory anemia with ring sideroblasts and marked thrombocytosis (RARS-T).2,4-6 This malignancy has been assigned as a provisional entity in the chapter “Myelodyplastic/myeloproliferative neoplasms, unclassifiable” of the WHO classification.7 Patients have anemia, clinical and morphological features of myelodysplastic syndromes (MDS), but also show marked thrombocytosis associated with abnormal megakaryocytes resembling BCR-ABL1 negative myeloproliferative neoplasms (MPN).7 In line with the myeloproliferative character of this disease, RARS-T patients often show JAK2V617F mutations and, much less common, MPLW515 mutations (MPLW515mut).7,8 To further characterize this entity, we analyzed 47 RARS-T patients for the occurrence of mutations in SF3B1 (SF3B1mut) and its association with JAK2V617F and MPLW515mut. To our knowledge, this is the largest cohort studied so far. All patients strictly met the criteria for RARS-T according to the WHO classification 2008.7 Twenty-seven of 47 (57.4%) were female and 20 of 47 (42.6%) were male. The median age was 76 years (range 44-89 years), white blood cell count (WBC) 7.9×10/L (range 2.9-60.0×10/L), hemoglobin level (Hb) 9.7 g/dL (range 6.9-13.1 g/dL), platelet count 691×10/L (range 466-1,500×10/L), and percentage of ring sideroblasts 60% (range 18-97%). Most patients had a normal karyotype (39 of 47, 83.0%) and 8 of 47 (17.0%) had various chromosomal aberrations consistent with MDS/MPN. Screening for SF3B1mut was performed by direct Sanger sequencing of exons 11-16. Using cDNA, a 1kb amplicon was generated with primers 5`-TGACCAGCCATCTGGAAATC-3` (forward, exon 10) and 5`-CACCATCTGTCCCACAACAC-3` (reverse, exon 17). Sequencing was performed with previous primers and 5`-GCTTGCCAGGACTTCTTGCT-3` (reverse, exon 14), 5`-AGCTTTTGCTGTTGTAGCCTCTG-3` (forward, exon 14). Mutation load was determined according to the ratio of mutated/wild-type in the electropherograms of forward and reverse reaction. Cases being SF3B1 unmutated by Sanger sequencing were reanalyzed with a more sensitive 454 deep-sequencing approach (NGS) on genomic DNA. The lowest detectable mutation load with NGS was about 3% (1,013 reads) and, for Sanger sequencing, 5-10% as confirmed by NGS. JAK2V617F, JAK2exon12 and MPLW515 mutations were analyzed as published previously.9-11 Sensitivities of the assays were approximately 1%, 10%, and 5%, respectively. The frequency of SF3B1mut was 41 of 47 (87.2%). The 6 patients without SF3B1mut were reanalyzed by NGS but, despite higher sensitivity, no SF3B1mut was found. In the 41 SF3B1mut patients, 43 mutations with median mutation load of 40% (range 15-70%) were detected. The most frequent mutation was Lys700Glu (22 of 43, 51.2%), followed by Lys666Arg/Asn/Thr (7 of 43, 16.3%), Glu622Asp (4 of 43, 9.3%), His662Gln (4 of 43, 9.3%), Arg625Lys/Cys (3 of 43, 7.0%), Thr627Pro, Met784_Lys785delinsIle and a synonymous Leu575Leu (1 of 43, 2.3% each). The two double mutated cases had: i) Arg625Cys (load 45%) and Thr627Pro (load 45%); or ii) Lys700Glu (load 30%) and the synonymous Leu575Leu (load 70%) mutation. JAK2V617F was also very frequent (36 of 47, 76.6%). Twenty-five of 36 (69.4%) revealed a mutation load of 10-50%, 6 of 36 (16.7%) had low (≤10%) and 5 of 36 (13.9%) high (>50%) mutation load. SF3B1mut and JAK2V617F were found concomitantly in 63.8% (30 of 47) patients. Remarkably, each of the 47 patients had at least either SF3B1mut or JAK2V617F (Figure 1). All JAK2V617F negative patients were additionally analyzed for JAK2exon12 but no mutation was detected. Only one patient (1 of 47, 2.1%) carried an MPLW515L in addition to SF3B1Lys700Glu. Age, Hb, WBC, platelet count, and percentage of ring sideroblasts were analyzed with t-test according to SF3B1mut and JAK2V617F. Platelet counts were significantly lower in JAK2wt compared to JAK2V617F patients (639 vs. 799×10/L, P=0.013). In contrast, SF3B1mut were associated with higher percentages of ring sideroblasts (SF3B1mut vs. SF3B1wt: 63% vs. 42%, P=0.016). SF3B1mut and JAK2V617F double mutated cases had significantly higher percentage of ring sideroblasts than all other cases (SF3B1mut/JAK2V617F vs. all others: 66 vs. 50%, P=0.011) (Table 1). However, there was no correlation between percentage of ring sideroblasts with SF3B1 mutation load (P=0.632). In accordance with the high SF3B1mut frequency in our cohort (87.2%), previous publications have also reported high frequencies of 66.7% (12 of 18),2 72.2% (13 of 18),4 83.3% (5 of 6)5 and 90.1% (10 of 11).6 Furthermore, we and others showed SF3B1mut to occur in combination with MPLW515mut and mainly JAK2V617F.2,4 Overlap of SF3B1mut and JAK2V617F were present in 63.8% patients, which is much more frequent than in previous studies with 11.1% (2 of 18)2 and 22.2% (4 of 18).4 This is mainly related to the high frequency of JAK2V617F (76.6%) in our RARS-T cohort, rarely seen in other studies (31-100%) in 3 to 23 analyzed patients.8 A very large recent multicenter study with 175 RARS-T patients analyzed for JAK2V617F reported a frequency of 42.9%.12 The only study of hematologic parameters in dependence of SF3B1mut in RARS-T patients detected no difference between SF3B1mut and SF3B1wt.2 Our results confirmed this finding except that SF3B1mut cases had a significantly higher percentage of ring sideroblasts. This is comparable to results seen in MDS patients.2 There is still ongoing discussion as to whether RARS-T is a distinct entity, or a subset of essential thrombocythemia or a progression of RARS.7,8 In contrast to MDS, almost no SF3B1mut were found in MPN patients,1,13 whereas JAK2V617F is characteristic for MPN and rare in MDS.12,14 Thus, the combination of high SF3B1mut and JAK2V617F frequencies supports the idea that RARS-T actually represents an overlap of MDS and MPN. Single RARS-T cases have been reported that developed from a pre-existing RARS through increasing JAK2V617F mutation load or the acquisition of JAK2V617F.12,15 As we could detect both RARS-T patients with high SF3B1mut and low JAK2V617F load and vice versa, our data support the idea that RARS-T might develop from two directions. In this process, MDS patients with SF3B1mut or MPN patients with JAK2V617F might each acquire the other mutation.

Table 1.Characteristics of patients according to SF3B1 and JAK2V617F mutations.

Figure 1.Frequencies and distribution of SF3B1, JAK2V617F and MPLW515 mutations. Each analyzed case is represented by a box. Mutated cases are col-ored in red, wild-type cases in gray.

In summary, our results show that RARS-T is characterized by high frequencies of SF3B1mut associated with percentages of ring sideroblasts and JAK2V617F correlated with platelet counts. Our data clearly demonstrate that the assignment of RARS-T into the WHO category “Myelodyplastic/myeloproliferative neoplasms” is not only justified based on cytomorphology, but is also reflected by molecular mutations. We strongly recommend screening for JAK2V617F and SF3B1mut in all confirmed or suspected RARS-T patients to facilitate diagnosis and distinguish RARS-T from other disease entities or reactive conditions.