Diagnostic and prognostic contribution of targeted NGS in patients with triple‐negative myeloproliferative neoplasms

Philadelphia negative myeloproliferative neoplasms (MPN Ph-) are clonal hematopoietic stem cell disorders characterized by proliferation in one or more myeloid cell lineages. Somatic mutations in JAK2, CALR and MPL genes currently constitute major diagnostic criteria according to the 2017 revised World Health Organization Classification.1 However, a subset (10-15%) of MPN Ph- patients, the so-called triple-negative (TN), do not present any of those driver mutations, but may harbor others that might be associated with inferior overall survival (OS) or leukemic transformation.2, 3 Based on this, and considering that there are just a few studies on this subgroup of patients, our aim was to assess the utility in terms of diagnosis and prognosis by performing targeted deep sequencing (TDS) of myeloid-related genes in TN MPN patients. A total of 43 TN patients that were consecutively referred to ICO-Badalona (Spain), from 2001 to 2012, were included in the study. All samples were collected with patient consent in accordance with the Declaration of Helsinki, after the approval by the local ethics committee (PI-15-001). According to 2017 WHO classification, the cohort included 35 patients with essential thrombocythemia (ET) and 8 with primary myelofibrosis (PMF). The TN patients were selected after excluding JAK2 V617F, CALR exon 9 and MPL exon 10 mutations by conventional techniques (allelic discrimination, DNA fragment analysis and Sanger sequencing, respectively). Genomic DNA was obtained either from bone marrow or peripheral blood samples, and was used for TDS studies. Barcoded libraries were prepared according to the manufacturer's instructions, using an amplicon based panel (HEAT-Seq Choice Target Enrichment System, Roche, Pleasanton, CA, USA) targeting frequently mutated regions of 17 myeloid-related genes4, 5 (Table S1). Samples were run on a MiSeq sequencer for paired-end 2x150 bp reads with a mean coverage of 850x. Analysis was performed using an in-house bioinformatic pipeline that included removal of PCR duplicates using molecular barcoding. Variants were filtered according to location (exonic and splicing), variant type (nonsynonymous, SNVs and indels), population frequency (minor allele frequency ≤ 0,01 according to dbSNP, ExAC, Exome Variant Server and 1000 Genomes project population databases) and variant allele frequency (VAF ≥5%). With TDS we detected a total of 20 variants among the studied cohort, with 15/43 (35%) previously classified as TN patients, harboring at least one somatic variant. Main clinical and biological data from the 15 mutated cases are summarized in Table 1. This group of patients was comprised by six males and nine females, with a mean age at diagnosis of 53 ± 20 years. Of them, 11 were ET patients while four corresponded to PMF cases. The MPL noncanonical variants (n = 5) and TET2 variants (n = 5) were the most frequent among patients. Three of the MPL variants correspond to the well-known S204P change located in exon 4, while the other 2 variants, E259K and V368 L, were in exons 5 and 7, respectively. Moreover, in an ET case, a single nucleotide deletion was detected in CALR gene in exon 9, which had not been previously detected by DNA fragment analysis technique. For this reason, we reclassified these six patients who had been previously misclassified as TN. Besides those genes, variants were also found in DNMT3A (n = 3), TP53 (n = 3), SF3B1 (n = 2) and EZH2 (n = 1). Most of the patients (n = 11) only showed one variant, but there were four cases with co-occurrence of up to three variants (Table 1). Median follow-up of living patients was 93 months (range: 7-343 months). Patients with PMF had an inferior OS compared to ET (10-year OS 40% IC95% 1-79 for PMF vs 10-year OS 83% IC95% 67-99 for ET, P = .011) and were characterized by a higher frequency of mutations (50% vs 31% of patients harboring at least one mutation), which is consistent with previous published studies.2, 5 Additionally, strictly considering real TN cases, no significant differences in OS were observed between the two groups of patients (mutated vs nonmutated cases) in the whole cohort (ET and PMF), whereas significant differences were observed in the small PMF cohort of patients (Figure S1). Some previous TDS-based studies failed to include all MPL exons in their panel, which could underestimate the real proportion of cases harboring a variant in this gene.2, 4, 5 It has been published that the percentages of patients with ET or PMF classified as TN but with non-driver mutations are 43% and 82%, respectively.2, 5 Therefore, the percentage of real TN cases among the MPN population may be lower than described. Of the three noncanonical MPL variants found in our study, S204P has been previously characterized by Cabagnols et al as a weak gain-of-function mutant which increases MPL signaling.6 Further studies are needed to characterize the other two variants detected in our cohort. Moreover, considering that these variants were not previously described in the literature, that we have not performed functional analysis, and that they were present in a VAF near 50%, neoplastic origin could not be ensured. This study also supports the fact that TN cases may present other mutations such as TET2, DNMT3A and SF3B1 that might have prognostic relevance and that may help to determine the clonal nature of the disease when a diagnosis is trying to be established, especially in patients with normal karyotype.1, 3 Although after considering only real TN cases we did not observe any difference in survival, prognostic relevance of non-driver mutations has been described in much larger cohorts of patients with PMF, ET and polycythemia vera.2, 5 To sum up, our panel has proven usefulness to detect variants at low and high VAFs in one-third of TN MPN cases, MPL and TET2 being the most frequent mutations found in this series. Even with a small cohort of patients, our data reinforces the idea that the inclusion of all MPL coding exons in TDS panels might be considered in order to cover possible noncanonical variants, which would be relevant for disease classification and prognostic risk stratification. Harboring one or more mutations may negatively influence OS. Studies in larger TN cohorts are needed in order to take conclusions regarding the impact of driver and non-driver mutations in this subset of patients. This work was supported in part by a grant from the Instituto de Salud Carlos III, Ministerio de Economia y Competividad, Spain (PI/17/0575); 2017 SGR288 (GRC) Generalitat de Catalunya, CERCA Programme/Generalitat de Catalunya, Fundació Internacional Josep Carreras, "la Caixa" Foundation and from Celgene Spain. Sequencing was performed at IGTP Genomic Core Facility (Badalona, Spain). Probes and reagents were provided by Roche Sequencing Solutions, Inc. (Pleasanton, CA, USA). Table S1 Genes with covered regions in the designed panel Figure S1. Overall survival curves considering only real triple-negative cases stratified according to the presence or absence of at least one somatic mutation. (A) ET and PMF cases altogether. (B) ET cases. (C) PMF cases Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.