Molecular abnormalities are prognostically relevant in morphological subtypes of myelodysplastic/myeloproliferative neoplasms (MDS/MPN), giving rise to contemporary molecularly integrated prognostic models.1-3 Established adverse prognostic associations include truncating ASXL1 mutations in chronic myelomonocytic leukemia (CMML)1 and MDS/MPN with ring sideroblasts and thrombocytosis (MDS/MPN-RS-T),3 TP53 and CBL mutations in unclassifiable MDS/MPN (MDS/MPN-U),4 and TET2 mutations in BCR-ABL1-negative atypical chronic myeloid leukemia.2 Recently, molecular signatures have been used to further stratify MDS/MPN-U patients into CMML-like (ASXL1, SRSF2, RUNX1, and/or NRAS mutant), MDS/MPN-RS-T-like (JAK2 and/or SF3B1 mutant), atypical chronic myeloid leukemia-like (SETBP1 and/or ASXL1 mutant), TP53 mutant and an “others” category.5,6 Despite their prognostic impact, these mutations are not specific for underlying disease entities. Recently, SF3B1 mutations were shown to be disease-defining in a subset of patients with MDS7,8 and CMML.9 Whether SF3B1 mutations are similarly diseasedefining in other myeloid subgroups is not known. Given the relative rarity of MDS/MPN patients, we assembled a large, molecularly annotated cohort of MDS/MPN patients to assess the clinical and prognostic impact of SF3B1 mutations, agnostic of disease morphology.
After Mayo Clinic institutional review board approval, clinical data from adult (age at diagnosis >18 years) patients with a World Health Organization (WHO)- defined diagnosis of MDS/MPN (CMML, MDS/MPN-U and MDS/MPN-RS-T), from 1994 to 2020, were included in the analysis. Patients with atypical chronic myeloid leukemia were excluded due to lack of uniform genetic annotation, limited SF3B1 mutations (n=2), and patient numbers (n<50). A separate cohort of SF3B1-mutant MDS patients diagnosed between 1994 to 2017 was included for comparison. An external cohort of patients from H. Lee Moffitt Cancer Center (Tampa, FL, USA) was used for independent validation after institutional review board approval. Next-generation sequencing for myeloid relevant genes was done at diagnosis or first referral, using institutional or commercially available myeloid malignancy-specific gene panels according to previously published methods.4 The distribution of continuous variables was statistically compared using nonparametric (Mann-Whitney or Kruskal-Wallis) tests, while nominal variables were compared using the c2 test. Time-to-event analyses (for overall [OS] and acute myeloid leukemiafree survival [LFS]) were performed using the method of Kaplan-Meier, with death (for OS), transformation to acute myeloid leukemia (for LFS), and allogeneic hematopoietic stem cell transplantation (for both OS and LFS) used as censors.
Overall, 778 consecutive WHO-defined MDS/MPN patients were included in the primary cohort (CMML, n=578 [74%]; MDS/MPN-RS-T, n=79 [10%] and MDS/MPN-U, n=121 [16%]). The median age was 72 (range, 18-95) years with 511 (66%) males (Table 1). Four (3%) patients in the MDS/MPN-U group met proposed criteria for oligomonocytic CMML and had an absolute monocyte count between 0.5 to 0.9 x 109/L, with monocytes constituting >10% of the total white blood cell count.10 Cytogenetic abnormalities (excluding sole -Y) were present in 197 (28%) of 695 assessable patients; 138 (70%) patients with a single karyotypic abnormality, 35 (18%) with a complex karyotype (defined as ≥3 independent structural/numerical abnormalities, excluding autosomal monosomies) and 26 (13%) with monosomal karyotypes, with frequent cytogenetic abnormalities including 51 (26%) +8, 49 (25%) -7/7q-, 23 (12%) 20q-, 12 (6%) 5q- (2 as sole abnormalities, classified as MDS/MPN based on morphology), 11 (6%) 13q-, 6 (3%) inv(3)/3q26 (3 GATA2-EVI1 fusion), and 5 (3%) with-11/11q23 (KMT2A). Cytogenetic risk stratification as per the CMML-specific scoring system (CPSS) cytogenetic stratification11 was predictive of OS (P<0.0001) in our cohort with 498 (72%) in the low-risk category (median OS 41 months [95% CI: 32-50]), 120 (17%) in the intermediate-risk category (median OS 21 months [95% CI: 16-33]), and 77 (11%) in the high-risk category (median OS 16 months [95% CI: 11-23]). Next-generation sequencing information at diagnosis was available for 444 (57%) patients with frequent molecular abnormalities being ASXL1 (n=235; 45%), SRSF2 (n=179; 40%), TET2 (n=155; 39%), SF3B1 (n=78, 15%), and DNMT3A (n=30, 7%) mutations (Table 1). At last median follow-up of 44 (95% CI: 37-50) months, transformation to acute myeloid leukemia had occurred in 123 (16%) patients, and 414 (53%) deaths had been documented. The Kaplan-Meier estimate of median OS was 32 (95% CI: 28-38) months (CMML 31 [95% CI: 27-37] months, MDS/MPN-RS-T 67 [95% CI: 43-101] months, and MDS/MPN-U 25 [95% CI 21-36] months), while the median was not reached for LFS. In the MDS/MPN cohort, there were 78 patients with SF3B1 mutations: 18 (23%) with CMML, 45 (58%) with MDS/MPN-RS-T, and 15 (19%) with MDS/MPN-U. There were 15 SF3B1 mutation hotspots (evaluable in 53 patients) with the most common abnormalities being K700E (n=24, 45%), H662Q (n=8, 15%), and K666R (n=6, 11%). The clinical and genomic characteristics are outlined in Online Supplementary Table S1.
We then combined all SF3B1-mutant MDS/MPN patients into one category (n=78) and compared them to their wild-type counterparts (n=446) (Table 1). The two groups had significant differences in clinical and molecular features as highlighted in Table 1. The median variant allele frequency (VAF) of mutant SF3B1 was 43% (range, 8-65) overall, being 43% (range, 8-65) in CMML patients, 43% (range, 12-50) in MDS/MPN-RS-T patients, and 40% (range, 16-52) in MDS/MPN-U patients (P=0.9), and was comparable to the median variant allele frequency of mutant ASXL1 at 37% (range, 11-52): CMML 37% (range, 27-37), MDS/MPN-RS-T 32% (range, 18-52), and MDS/MPN-U 29% (range, 11-43). As expected, a higher frequency of SF3B1-mutant versus SF3B1-wild type MDS/MPN patients (21% vs. 2%, P<0.0001) were treated with lenalidomide and erythropoiesis-stimulating agents (64% vs. 39%, P<0.0001), but the frequency of hypomethylating agent therapy use was similar (21% vs. 32%, P=0.1) (Table 1). The SF3B1 mutant cohort had a lower rate of transformation to acute myeloid leukemia (5% vs. 18%, P=0.0006) in comparison to the SF3B1-wild type cohort. The Kaplan-Meier estimates of LFS (median not reached in both groups, P=0.0002) and OS (57 vs. 31 months, P=0.03) were higher in the SF3B1-mutant MDS/MPN patients (Table 1 and Figure 1). These findings were validated in an external MDS/MPN cohort from Moffitt Cancer Center comprising 380 patients, 253 with CMML, 80 with MDS/MPN-RS-T, and 47 with MDS/MPN-U. The validation cohort was similar to the Mayo Clinic cohort in terms of age (P=0.4) and median follow-up (P=0.1). Importantly, SF3B1-mutant VAF was not predictive of OS in either the Mayo Clinic cohort (P=0.3) or the Moffitt Cancer Center cohort (P=0.7). In addition, there were no differences in OS between patients with mutations in the K700E hotspot and non- K700E sites in either the Mayo Clinic cohort (median OS 49 [95% CI: 22-109] months vs. 67 [95% CI: 36-126] months, P=0.5) (Online Supplementary Figure S1A) or the Moffitt Cancer Center cohort (median OS 85 [95% CI) months vs. not reached, P=0.9) (Online Supplementary Figure S1B).
We then compared SF3B1-mutant MDS/MPN patients (n=78) with SF3B1-mutant MDS patients (n=75) (Table 1). SF3B1-mutant MDS/MPN patients had a higher frequency of JAK2 V617F mutations (25% vs. 1%, P<0.0001; 10% vs. 1%, P=0.002 when MDS/MPN-RS-T patients were excluded) (Figure 2A, Table 1). When the SF3B1 mutation hotspots were compared between the two groups, SF3B1 K700E was the most common hotspot in both categories and was present in 24 (47%) MDS/MPN patients and 39 (53%) SF3B1-mutant MDS patients (P=0.5) (Online Supplementary Figure S1C, D). Overall, there were seven patients with co-occurring SF3B1 and SRSF2 mutations (4 with CMML, 3 with SF3B1-mutant MDS). Mutation details were available for two CMML patients; SF3B1 Y623C (42%)/SRSF2 P95H (45%) and SF3B1 K700E (45.2%)/SRSF2 P95H (2.8%), and two SF3B1-mutant MDS patients; SF3B1 K666Q (29%)/SRSF2 P95T (48%) and SF3B1 K700E (9%)/SRSF2 P95R (29%). At last median follow-up of 102 (95% CI: 63-141) months, there were no significant differences in rates of transformation to acute myeloid leukemia (5% vs. 3%, P=0.4), Kaplan-Meier estimates of median LFS (median not reached, P=0.3) or median OS (median, 57 vs. 65 months, P=0.2) between the two cohorts (Figure 2B, Table 1).
We then stratified SF3B1-mutant MDS/MPN patients by morphological features such as percentage of ring sideroblasts in bone marrow and percentages of blasts in peripheral blood and bone marrow. In a univariate survival analysis, percentage of peripheral blood blasts (P=0.1), percentage of bone marrow blasts ≥5 (P=0.4), abnormal karyotype (P=0.3), revised International Prognostic Scoring System score (P=0.8) and CPSS cytogenetic group (P=0.5) were not predictive of OS. Molecular abnormalities (overall frequency ≥5%) such as ASXL1 (P=0.3), TET2 (P=0.08), DNMT3A (P=0.6), JAK2 V617F (P=0.8), U2AF1 (P=0.2), SRSF2 (P=0.7), ZRSR2 (P=0.3), CBL (P=0.3) NRAS (P=0.8), or any RAS pathway mutation (KRAS/NRAS/CBL/PTPN11, P=0.3) did not affect OS (only 1 patient each had TP53 and RUNX1 mutations). Additionally, WHO criteria were unable to prognostically distinguish both Mayo Clinic (P=0.3) and combined (Mayo Clinic and Moffitt Cancer Center) cohorts of SF3B1-mutant MDS/MPN patients (P=0.7). Furthermore, neither the standard International Prognostic Scoring System (P=0.3), nor the revised version (P=0.7) was able to stratify SF3B1-mutant and MDS/MPN patients into prognostically relevant subtypes.
Finally, we conducted a multivariate analysis in the combined Mayo Clinic cohort of MDS/MPN and MDS patients with known independent prognostic factors in myeloid malignancies such as hemoglobin <10 g/dL, age ≥70 years, platelet count ≥450 x 109/L, cytogenetic subtypes (as per CPSS stratification), SF3B1 and ASXL1 mutations, and bone marrow blast percentage ≥5, and found that SF3B1 mutations retained their independent favorable prognostic impact (P=0.01) (Online Supplementary Table S2).
In summary, our data indicate that SF3B1-mutant MDS/MPN is a clinically and genomically distinct category within overlap myeloid neoplasms and, pending further validation, should be considered as a unique prognostic entity. Additionally, patients with this condition have distinct clinical and molecular characteristics in comparison to SF3B1-mutant MDS patients, arguing against a uniform classification category of SF3B1- mutant myeloid neoplasms.
Limitations of our study include smaller numbers of patients in certain subgroup comparisons, differential follow- up times and therapy choices, and selection biases largely due to the retrospective nature of the analyses.
- Received December 7, 2021
- Accepted January 28, 2022
Disclosures: AM has received research funding from BMS. EP has received honoraria from and/or serves on advisory boards for BluePrint, CTI, Stemline Therapeutics, Taiho and BMS; and has received research funding from Kura, Incyte, and BMS. MP has received research funding from Kura Oncology and Stemline Pharmaceuticals. All other authors declare that they have no conflicts of interest.
Contributions: AM compiled the clinical and genomics data, analyzed data, and wrote all the drafts of the manuscript. TL and CF performed the genomics analysis. RPK and KKR reviewed the pathology information and edited all drafts of the manuscript. KM, NG, AAK, KHB, WHH, MRL, HA, MS, AP, and AT contributed patients and edited all drafts of the manuscript. NHA, CT, DS, EP and RK contributed data from the external (validation) cohort and edited all drafts of the manuscript. MP conceptualized the study and edited all drafts of the manuscript. All authors contributed to the writing of the manuscript.
Data-sharing statement: for original data, please contact
this publication was supported by grant #UL1 TR002377 from the National Center for Advancing Translational Sciences (NCATS); however, the findings do not necessarily represent official views of the National Institutes of Health.
- Patnaik MM, Itzykson R, Lasho T. ASXL1 and SETBP1 mutations and their prognostic contribution in chronic myelomonocytic leukemia: a two-center study of 466 patients. Leukemia. 2014; 28(11):2206. https://doi.org/10.1038/leu.2014.125PubMedGoogle Scholar
- Patnaik MM, Barraco D, Lasho TL. Targeted next generation sequencing and identification of risk factors in World Health Organization defined atypical chronic myeloid leukemia. Am J Hematol. 2017; 92(6):542-548. https://doi.org/10.1002/ajh.24722PubMedPubMed CentralGoogle Scholar
- Patnaik MM, Lasho TL, Finke CM. Predictors of survival in refractory anemia with ring sideroblasts and thrombocytosis (RARST) and the role of next-generation sequencing. Am J Hematol. 2016; 91(5):492-498. https://doi.org/10.1002/ajh.24332PubMedGoogle Scholar
- Mangaonkar AA, Swoboda DM, Coltro G. Clinicopathologic characteristics, prognostication and treatment outcomes for myelodysplastic/myeloproliferative neoplasm, unclassifiable (MDS/MPN-U): Mayo Clinic-Moffitt Cancer Center study of 135 consecutive patients. Leukemia. 2020; 34(2):656-661. https://doi.org/10.1038/s41375-019-0574-xPubMedGoogle Scholar
- Palomo L, Meggendorfer M, Hutter S. Molecular landscape and clonal architecture of adult myelodysplastic/myeloproliferative neoplasms. Blood. 2020; 136(16):1851-1862. https://doi.org/10.1182/blood.2019004229PubMedPubMed CentralGoogle Scholar
- Mangaonkar AA, Swoboda DM, Lasho TL. Genomic stratification of myelodysplastic/myeloproliferative neoplasms, unclassifiable: sorting through the unsorted. Leukemia. 2021; 35(11):3329-3333. https://doi.org/10.1038/s41375-021-01258-6PubMedGoogle Scholar
- Malcovati L, Karimi M, Papaemmanuil E. SF3B1 mutation identifies a distinct subset of myelodysplastic syndrome with ring sideroblasts. Blood. 2015; 126(2):233-241. https://doi.org/10.1182/blood-2015-03-633537PubMedPubMed CentralGoogle Scholar
- Malcovati L, Stevenson K, Papaemmanuil E. SF3B1-mutant MDS as a distinct disease subtype: a proposal from the International Working Group for the Prognosis of MDS. Blood. 2020; 136(2):157-170. https://doi.org/10.1182/blood.2020004850PubMedPubMed CentralGoogle Scholar
- Wudhikarn K, Loghavi S, Mangaonkar AA. SF3B1-mutant CMML defines a predominantly dysplastic CMML subtype with a superior acute leukemia-free survival. Blood Adv. 2020; 4(22):5716-5721. Google Scholar
- Valent P, Orazi A, Savona MR. Proposed diagnostic criteria for classical chronic myelomonocytic leukemia (CMML), CMML variants and pre-CMML conditions. Haematologica. 2019; 104(10):1935-1949. https://doi.org/10.3324/haematol.2019.222059PubMedPubMed CentralGoogle Scholar
- Such E, Germing U, Malcovati L, Hrodek O. Development and validation of a prognostic scoring system for patients with chronic myelomonocytic leukemia. Transfuze a Hematologie Dnes. 2013; 19(3):191. Google Scholar
Figures & Tables
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.