Annals of Hematology (2019) 98:2319–2328 https://doi.org/10.1007/s00277-019-03766-z ORIGINAL ARTICLE Genomic characterization in triple-negative primary myelofibrosis and other myeloid neoplasms with bone marrow fibrosis Alberto Alvarez-Larrán1 & Mónica López-Guerra2,3 & María Rozman2 & Juan-Gonzalo Correa1 & Juan Carlos Hernández-Boluda4 & Mar Tormo4 & Daniel Martínez2 & Iván Martín4 & Dolors Colomer2,3 & Jordi Esteve1 & Francisco Cervantes1 Received: 3 June 2019 /Accepted: 20 July 2019 /Published online: 8 August 2019 # Springer-Verlag GmbH Germany, part of Springer Nature 2019 Abstract Triple-negative primary myelofibrosis (TN-PMF) and other myeloid neoplasms with associated bone marrow fibrosis such as the myelodysplastic syndromes (MDS-F) or the myelodysplastic/myeloproliferative neoplasms (MDS/MPN-F) are rare entities, often difficult to distinguish from each other. Thirty-four patients previously diagnosed with TN-PMF (n = 14), MDS-F (n = 18), or MDS/MPN-F (n = 2) were included in the present study. After central revision of the bone marrow histology, diagnoses according to the 2016-WHO classification were TN-PMF (n = 6), MDS-F (n =19),andMDS/MPN-F(n = 9), with TN-PMF genotype representing only 4% of a cohort of 141 molecularly annotated PMF. Genomic classification according to next- generation sequencing and cytogenetic study was performed in 28 cases. Median number of mutations was 4 (range 1–7) in cases with TP53 disruption/aneuploidy or with chromatin-spliceosome mutations versus 1 mutation (range 0–2) in other molec- ular subgroups (p < 0.0001). The number of mutations and the molecular classification were better than PMF and MDS con- ventional scoring systems to predict survival and progression to acute leukemia. In conclusion, TN-PMF is an uncommon entity when the 2016 WHO criteria are strictly applied. Genomic classification may help in the prognostic assessment of patients with myeloid neoplasms with bone marrow fibrosis. Keywords Primary myelofibrosis . Myelodysplastic syndromes . Diagnosis . Prognosis . Genomic classification Introduction Primary myelofibrosis (PMF) is the less frequent of the classic myeloproliferative neoplasms (MPN) and the one with the worst prognosis, with an estimated median survival around Mónica López-Guerra and María Rozman contributed equally to this 6 years in contemporary series [1–4]. The majority of PMF work. patients have mutations in JAK2V617F, CALR,orMPL genes. Electronic supplementary material The online version of this article However, up to 10% of them are negative for these mutations, (https://doi.org/10.1007/s00277-019-03766-z) contains supplementary with these triple-negative PMF (TN-PMF) patients showing a material, which is available to authorized users. more unfavorable prognosis [2–4]. In such cases, the absence * Alberto Alvarez-Larrán of a driver mutation can often complicate their distinction [email protected] from other myeloid neoplasms with associated bone marrow fibrosis. Myelodysplastic syndrome with associated fibrosis (MDS- 1 Hematology Department, Hospital Clínic, IDIBAPS, Villarroel 170, 08036 Barcelona, Spain F) represents 10–20% of MDS and, despite not being recog- nized as a separate entity, it is associated with an unfavorable 2 Hematopathology Section, Pathology Department, Hospital Clínic, IDIBAPS, Barcelona, Spain outcome [5– 7]. In addition, myelodysplastic/ myeloproliferative neoplasms (MDS/MPN) are rare entities 3 CIBERONC, Salamanca, Spain 4 sharing clinical and morphological features with MDS and Hematology Department, Hospital Clínico-INCLIVA, MPN and including chronic myelomonocytic leukemia Valencia, Spain 2320 Ann Hematol (2019) 98:2319–2328 (CMML), atypical BCR-ABL1-negative chronic myeloid leu- paratrabecular localization, megakaryocyte size (small kemia (aCML), MDS/MPN with ring sideroblasts and and/or giant), and nuclear features, including bulbous thrombocytosis (MDS/MPN-RS-T), and MDS/MPN unclas- morphology, hypolobulation with disperse nuclei, sified (MDS/MPN-U) [1]. In all these entities, if bone marrow hyperlobulation (stag horn-like), and hyperchromatism, fibrosis is present, differential diagnosis with TN-PMF is were also evaluated. Taking into consideration the challenging. above-mentioned features, all cases were finally classified It has recently been shown that a significant proportion according to the current WHO criteria [1]. of patients with MDS and MPN have mutations in a large Twenty-seven cases with available samples for NGS variety of genes associated with myeloid neoplasms [8, study were analyzed. Targeted NGS was performed using 9]. In addition, the presence of mutations in certain genes the Sophia Genetics Myeloid Tumor Solution Panel in- and the number of mutated genes has been associated with cluding the following genes: ABL1, ASXL1, BRAF, a more unfavorable prognosis in both MDS and PMF CALR, CBL, CEBPA, CSF3R, CSNK1A1, DNMT3A, [8–13]. ETV6, EZH2, FLT3, HRAS, IDH1, IDH2, JAK2, KIT, In the present study, we have characterized by next- KMT2A, KRAS, MPL, NPM1, NRAS, PTPN11, RUNX1, generation sequencing (NGS) techniques a cohort of patients SETBP1, SF3B1, SRSF2, TET2, TP53, U2AF1, U2AF2, homogeneously diagnosed with TN-PMF, MDS-F, or MDS/ WT1,andZRSR2. Libraries were sequenced 2 × 300 bp MPN-F according to the 2016 WHO criteria. The aim of the on an Illumina MiSeq. Sequencing data were analyzed study was to evaluate if mutations detected by NGS can help using the commercial software Sophia DDM v4.4 in establishing a definite diagnosis among these entities or (Sophia Genetics). Only variants categorized as highly refining the prognostic assessment of the patients. or potentially pathogenic by Sophia software were consid- ered. The minimum variant allele frequency (VAF) con- sidered was 2%. Synonymous, intronic, and polymorphic Patients and methods variants were discarded. Additionally, selected variants were reviewed in COSMIC and ClinVar databases to se- A search for TN-PMF and other chronic myeloid neoplasms lect clinically relevant mutations. with bone marrow fibrosis was performed in two Spanish Genomic classification combining cytogenetics and institutions. Cases with CMML, aCML, MDS/MPN-RS-T, NGS studies was performed as reported by Grinfeld and MDS with excess blasts type-2 were excluded. Initial et al. For such purpose, patients were hierarchically allo- data, including the main clinical, hematological, and cytoge- cated into eight molecular subgroups as follows: group 1: netic features, as well as the therapies administered during TP53 disruption or aneuploidy (TP53 mutation, follow-up, transformation to acute leukemia, and the causes Chr17pLOH or Chr5-/Chr5q-), group 2: ≥ 1 chromatin of death were collected. or spliceosome mutation (EZH2, IDH1, IDH2, ASXL1, Bone marrow biopsies were performed at diagnosis in PHF6, CUX1, ZRSR2, SRSF2, U2AF1, KRAS, NRAS, all cases. Hematoxylin/eosin and reticulin stained slides GNAS, CBL, Chr7/7qLOH, Chr4q/LOH, RUNX1, and CD34 immunostaining when available of formalin- STAG2,andBCOR), group 3: CALR mutation, group 4: fixed or Bouin’s solution fixed, paraffin embedded tre- MPL mutation, group 5: homozygous JAK2 mutation, phine biopsies were centrally reviewed by two expert group 6: heterozygous JAK2 mutation, group 7: myeloid hematopathologists. The review was performed blindly neoplasm with other driver mutation, and group 8: mye- without knowledge of age, clinical symptoms, original loid neoplasm with no known driver mutation [9]. diagnosis, or gene mutational status. Histological features The clinical and biological data at diagnosis, the cyto- assessed were cellularity (measured as percentage), pres- genetic findings, and the results from NGS analysis were ence of lymphoid aggregates (yes or no), granulocytic included in the statistical analysis. Comparisons among precursors (increased, normal or decreased, left shifted), diagnostic groups were performed after central review of erythrocytic precursors (increased, normal or decreased, bone marrow histology. The prognostic group was calcu- left shifted, megaloblastoid changes), myeloid/erythroid lated according to the International Prognostic Scoring ratio (increased, normal or decreased), presence of sinu- Systems (IPSS) designed for PMF [15]andMDS[16], soidal hyperplasia (yes or no), presence of intrasinusoidal respectively. The probabilities of survival and progression hematopoiesis (yes or no), and grade of reticulin fibrosis to acute leukemia were estimated by the Kaplan Meier according to the European Consensus Grading system method using the log rank test for comparisons. (ranging from MF-0 to MF-3) [14]. An extensive quanti- Informed consent for the scientific use of the patients’ fication and description of the morphology and distribu- clinico-hematological data and biological samples was ob- tion of the megakaryocytes was performed. Presence of tained, and the study was approved by the Ethics dense and loose clusters of megakaryocytes, Committee of the Hospital Clínic of Barcelona. Ann Hematol (2019) 98:2319–2328 2321 Table 1 Main clinical and hematological characteristics at TN-PMF MDS-F MDS/MPN-U-F P value diagnosis in 34 patients with N =6 N =19 N =9 myeloid neoplasms with bone marrow fibrosis Age, year* 61 (48–78) 64 (35–80) 65 (42–79) ns Male sex, n (%) 3 (50) 12 (63) 8 (89) ns Palpable spleen, n (%) 3/6 (50) 6/18 (33) 8/9 (89) 0.02 WBC count, × 109/L* 10.1 (3.8–19) 3 (0.3–11.4) 15.4 (5.1–69) < 0.001 Hemoglobin, g/L* 88 (78–144) 88 (69–140) 110 (50–140) ns Platelets, × 109/L* 241 (26–600) 85 (4–465)