Study of mutation complexity in Chronic Myeloproliferative Neoplasms: pathogenetic insights and translational relevance

The classic chronic myeloproliferative neoplasms (MPNs) comprise polycythemia vera (PV), essential thrombocythemia (ET) and primary myelofibrosis (PMF), originate from de-regulated clonal proliferation of hematopoietic stem cells and are associated with overproduction of mature blood cells. MPN are characterized by a high molecular complexity. Although JAK2 mutations have been shown to be the phenotypic drivers in MPN, there is evidence of clonality and mutational events preceding the acquisition of JAK2V617F. An increasing number of mutations in genes distinct from JAK2 have been recently identified in patients with MPN. Mutations can be divided in: “phenotypic driver” (JAK2, MPL and CALR) since the expression of the mutated gene in cell lines caused cytokine independent growth and in animal models phenotypes closely resembling a myeloproliferative disease; “subclonal mutations” that usually occur in hematopoietic cell subclones of variable size, often but not invariably together with one of the phenotypic driver mutations. Although different studies have found statistically significant correlations between the presence (or allelic burden) of certain molecular abnormalities and clinical phenotype in MPN, particularly in myelofibrosis, these novel molecular information still have modest influence on risk stratification and treatment decisions. This project focused on identifying novel disease-associated genetic alterations in MPNs, addressing their relevance for disease progression and leukemia transformation. Elucidation of these issues would facilitate the understanding of genotype-phenotype correlations, in order to explain the molecular puzzle underlying the pathogenesis of MPNs, as well as refining risk stratification and improving therapeutic management of the patients. In the first set of experiments we studied two independent cohorts of patients with PMF, in order to clarify the prognostic relevance of newly described mutations, including TET2, CBL, IDH1 or IDH2, ASXL1, EZH2, DNMT3A and SRSF2. The first cohort of 483 patients, from multiple centres in Europe, studied within one year of diagnosis, was needful to select out which one of these mutations predicted worse outcome. Overall and leukemia-free survivals (OS and LFS) were inter-independently predicted by ASXL1, SRSF2 or EZH2 mutations and ASXL1, SRSF2 and IDH1/2 mutations, respectively. We classified the entire patient cohort into those who displayed at least one (“molecularly high-risk”, HMR) or (“molecularly low risk”, LMR) none of the four mutations and we found that pts at LMR had better outcome respect to pts in HMR category. The correlation between the mutational status and the disease outcome was significant in terms of both OS and LFS. We then found that also the “number” of prognostically detrimental mutated genes had impact on outcome. Patients with 2 or more HMR mutations had significantly reduced overall and leukemia-free survival compared no only to unmutated but also to those presenting only one mutation in HMR genes. These results were validated both in learning and validation cohorts. In the second part of my work I focused my attention on the very recent discovery of CALR mutations in MPNs in order to clarify different aspects regarding the identification and the impact of these mutations on clinics. Based on the common feature of CALR mutations to generate the same frame at C-terminus of the protein, we developed a polyclonal antibody against a 17 peptide residue in the variant C-terminus of mutated calreticulin, that proved able to label selectively bone marrow sections from CALR mutated patients, distinguishing them from normal subjects and ET and PMF patients harboring the JAK2 or MPL mutations. The immunostaining was almost entirely confined to cells of the megakaryocyte lineage, while myeloid and erythroid cells showed fainter labelling. So we demonstrated that the use of mutation specific antibody may be useful for molecular classification of ET and PMF patients and also for studying different aspects of hematopoiesis, and particularly megakaryocytopoiesis, in patients harboring CALR mutations. In a second moment we investigated the prevalence, characteristics and clinical and laboratory features associated with CALR mutations and the prognostic relevance of mutated CALR in large series of ET and PMF patients, in multicentre studies, and we demonstrated that CALR-mutant ET and CALR-mutant PMF have a relatively indolent clinical course compared with the respective JAK2-mutant and MPL-mutant disorders. On the contrary, PMF with non-mutated JAK2, CALR, or MPL (triple negative, TN) has a poor prognosis with a particularly high risk of leukemic transformation. Also the prognostic advantage of CALR mutation in PMF regards only patients harboring type 1/type 1-like mutation, while the survival of those harboring type 2/type 2-like mutation does not differ from JAK2V617F and MPLW515 mutated ones, maybe due to the molecular complexity of subclonal mutations. The interesting results that we obtained in the study on primary myelofibrosis, in terms of better OS of CALR mutated patients compared with JAK2+, MPL+ and TN pts, and the association of high molecular risk signature with negative outcome in PMF patients, prompt us at defining the role for outcome prediction of these mutational status in secondary myelofibrosis (sMF). We evaluated Post-Polycythemia Vera and Post-Essential Thrombocythemia Myelofibrosis (PPV-MF and PET-MF) patients (from 4 Italian centres) at time of diagnosis and we reported that in sMF there is not a strong impact of prognostically detrimental mutated genes on outcome that we conversely demonstrated in PMF patients, even if ASXL1 and SRSF2 still negative prognostic factor in sMF. We can suppose that in the secondary form the mutational profiling is much complex, probably due to the acquisition of different mutations, at different point during the disease progression. In our study on ET and PMF triple negative resulted the category with the worst prognosis, they present a higher risk of developing anemia and thrombocytopenia, suffer from reduced overall survival compared to other genotypes, particularly to CALR type1/type1like mutations, and may be at higher risk of leukemic transformation. Therefore, in the last part of my work, I focused my attention on an extensive molecular characterization of 28 PMF resulted triple negative in our cohort of 396 PMF patients. We evaluated 18 genes known to be prognostically relevant in previous studies (EZH2, ASXL1, IDH1/2, SRSF2, TP53, TET2, RUNX1, CBL, NRAS, KRAS, DNMT3A, SF3B1, IKZF1, NFE2, SH2B3, U2AF1). We observed that among triple negative patients there was a significantly higher proportion of patients clustered in the high molecular risk (both with at least 1 and >2 HMR mutations), compared with JAK2V617F, CALR Type1/1-like and CALR Type2-type2-like mutated patients. We could observe that patients displaying more than 2 mutations had the worst outcome, both in terms of overall and leukemia-free survival, but without reaching statistical significance. So we can conclude that the outcome of triple-negative PMF is markedly worse than JAK2, CALR or MPL mutant PMF, the dismal outcome could be explained by the high number of subclonal mutations displayed in these patients. Overall, these findings show that the molecular landscape of MPN is even more complex and that mutational profiling, in particular in PMF, is needful for refine current risk stratification models and highly relevant to clinical decision-making as regards diagnostic approach and prognostication.