Genomic trajectory in leukemogenesis of myeloproliferative neoplasms: a case report

We report a patient with Essential Thrombocythemia (ET), subsequently diagnosed with concurrent myeloid and lymphoid leukemia. Generally, the molecular mechanisms underlying leukemic transformation of Philadelphia-negative myeloproliferative neoplasms (Ph-MPN) are poorly understood. Risk of transformation to acute myelogenous leukemia (AML) is low; transformation to both AML and acute lymphoblastic leukemia (ALL) is extremely low. Genetic defects, including allele burden, order of mutation acquisition, clonal heterogeneity and epigenetic mechanisms are important contributors to disease acceleration. A 78-year-old Caucasian female originally treated for stable ET, underwent disease acceleration and transition to myeloid sarcoma and B-cell ALL. Genomic reconstruction based on targeted sequencing revealed the presence of a large del(5q) in all three malignancies and somatic driver mutations: TET2, TP53, SF3B1, and ASXL1 at high allele frequency. We propose that the combination of genetic and molecular abnormalities led to hematopoietic stem cell (HSC) injury and disease progression through sub-clone branching. We hypothesize that ancestral reconstruction of genomic data is a useful tool to uncover subclonal events leading to transformation. The use of ancestral reconstruction of genomic data sheds light on the unique clinical scenario described in this case report. By determining the mutational profile of tumors at several timepoints and deducing the most parsimonious relationship between them, we propose a reconstruction of their origin. We propose that blast progression originated from subclonal events with malignant potential, which coexisted with but did not originate from JAK2 p.V617F-positive ET. We conclude that the application of genomic reconstruction enhances our understanding of leukemogenesis by identifying the timing of molecular events, potentially leading to better chemotherapy choices as well as the development of new targeted therapies.


Background
The molecular mechanisms underlying leukemic transformation of Philadelphia-negative myeloproliferative neoplasms (Ph-MPN) are poorly understood. Risk of transformation to acute myelogenous leukemia (AML) is approximately 10-20% for patients with primary myelofibrosis (PMF) and considerably lower for those with Polycythemia Vera (PV) and Essential Thrombocythemia (ET), 5-10% and 2-5% respectively [1,2]. Transformation of MPN to acute lymphoblastic leukemia (ALL) is rare, with only seventeen cases reported in the literature [3]. Retrospective studies suggest that genetic defects, including allele burden, order of mutation acquisition, clonal heterogeneity and epigenetic mechanisms play an important role in the observed conversion rate, as well as the varied clinical-pathologic entities that evolve [4,5].

Open Access
*Correspondence: iberia.sosa@fccc.edu † Yujie Chen, Rafee Talukder contributed equally to this work 4 Fox Chase Cancer Center, 333 Cottman Ave, Philadelphia, PA 19111, USA Full list of author information is available at the end of the article In this report, we describe a series of clonal events in an elderly patient originally diagnosed with JAK2 V617Fpositive ET who presented in accelerated phase, with subsequent progression to concurrent myeloid sarcoma (MS) and B-cell ALL. We hypothesize that reconstruction of genomic mutations to build an ancestral tree may be a useful tool to characterize leukemogenesis in patients with high-risk disease.

Case presentation
A 78-year-old Caucasian female was diagnosed with JAK2 V617F-positive ET in 2010 based on laboratory data and molecular profiling of peripheral blood. She was initiated on hydroxyurea (HU) and low-dose aspirin. The patient remained on anticoagulation with coumadin due to a prior history of atrial fibrillation. She demonstrated stable and adequate platelet response to 10 mg/kg of HU.
In January 2018, she presented with shortness of breath, abdominal pain, nausea and vomiting. Computed tomography (CT) scan of chest, abdomen and pelvis revealed a large 7.9 cm × 4.9 cm ileocecal mass inducing partial colonic obstruction and moderate left pleural effusion. Pathology of the excised cecal mass showed a cluster of medium-sized MPO-positive myeloblasts with eccentric nuclei and scant cytoplasm, positive for CD33 and CD117 based on flow cytometry and consistent with MS (Fig. 1b). Macroscopic examination of her pleural fluid revealed medium-sized cells with large nuclei and scant basophilic cytoplasm positive for CD19, TdT and CD20 consistent with B-cell ALL (Fig. 1c). NGS of both the colonic mass and pleural fluid revealed SF3B1, TP53 exon 6 and exon 3 and TET2 (Table 1). JAK2-V617F and MPL were negative in both the colonic MS and B-cell ALL (Fig. 2).
Amplicon-based NGS of a panel of 48 genes commonly mutated in hematologic malignancies was An inferred phylogenetic tree was constructed from NGS and karyotype data from bone marrow samples obtained during the evolution from ET to MDS, as well as from colon and pleural effusion samples obtained during blast transformation. The corresponding timescale data points were aligned to reflect the hypothetical evolution of hematopoietic stem cells (HSC) most parsimonious with the sequencing and FISH findings reflected in Table 1. Note that between any two splits of the tree, the order of mutations and chromosomal deletions cannot be determined, and mutations are depicted sequentially for visual convenience (Fig. 2).

Discussion and Conclusions
In this single case report, genomic reconstruction importantly revealed the presence of a large del(5q) in all three malignancies: ET, MS and B-cell ALL, suggesting a common cell of origin. In chronic phase MPN, this finding should inform physicians about the imminent likelihood for disease progression. Episomal reprogramming has identified del(5q) as an early cytogenetic lesion with the capacity to perturb genome stability and differentiation [6]. Larger del(5q) size has been correlated with higher mutation frequency [5], which in this case included the somatic driver mutations: TET2, TP53, SF3B1, and ASXL1 at high allele frequency (Table 1). We propose that this combination led to HSC injury and disease progression through sub-clone branching (Fig. 2). The shared presence of JAK2-V617F and TET2 mutations in ET suggests that "megakaryocytic branching" originated directly from the HSC as opposed to a more lineage-restricted progenitor [7]. The JAK2-V617F mutation confers a weak proliferative advantage to HSC and its absence in blast phases suggests MS and B-cell ALL did not emerge from this subclone. Before blast conversion, the patient received lenalidomide, an immune modulator that yields cytogenetic remission by inhibiting growth of del(5q) progenitors without affecting other cells [4]. Clinically, lenalidomide improves survival and reduces transfusion requirements in patients with del(5q) MDS [4,8]. Some reports suggest an association between lenalidomide therapy and transformation to more aggressive phenotypes, including a review documenting transformation of MPN to ALL [3]. Emerging data suggest that progression to AML in patients treated with lenalidomide is associated with karyotype complexity and clonal selection rather than a drug-mediated transformation [4,9]. Accordingly, in our case, bone marrow biopsies after lenalidomide therapy reveal absence of del(5q) clones, likely as a result of known suppression from drug. They also revealed a new 20q deletion (Table 1), which may portend the malignant potential of uninhibited clones, as evidenced later by its high expression in B-cell ALL.
Based on the ancestral reconstruction of genomic data modeled in Fig. 2, we hypothesize a mechanism for disease acceleration, whereby subclonal events with potential for blast conversion coexisted with but did not originate from JAK2 p.V617F-positive ET. The presence of del(5q) and molecular abnormalities (TP53, KMT2A) in both MS and B-cell ALL (Fig. 2) suggest that "disease progression" originated from diversification of a pluripotent HSC capable of both myeloid and lymphoid differentiation [10,11], thereby leading to two distinct leukemia initiating cells (LIC): one containing 17p deletion and the other 20q deletion. Interestingly, del(5q) was suppressed in bone marrow but not from extramedullary sites where the blast phase manifested, suggesting variable sensitivity of different clones to lenalidomide. The concurrent presence of TP53 mutation in extramedullary sites [12] is known to confer a negative impact on survival and drug response to patients with del(5q) MDS treated with lenalidomide [7,13]. Recent studies suggest that patients with high-risk MDS, characterized by unfavorablerisk cytogenetic abnormalities and/or TP53 mutations, exhibit favorable clinical responses with robust mutation clearance when treated with hypomethylating agents (HMA) [14]. Unfortunately, HMA do not provide durable responses. A combination of azacytidine plus anti-CD47 monoclonal antibody is currently being investigated on TP53-AML with preliminary results showing an objective response in 71% of subjects and 48% complete remission [15].
We are limited in our ability to confirm the proposed order of pathogenic mutations. We lack banked bone marrow cells at all time points of disease evolution to demonstrate the proposed patterns of clonal progression. As this is a single case report, computer simulation of similar cases of accelerated phase ET would reaffirm our proposed model. However, collecting replicate cases is difficult given the rarity of the events described. Despite the limitations in our analysis, the denote loss of mutation. The order of mutations between branch splits cannot be inferred from the data and therefore they are listed lexicographically. The founder cell is proposed to be an HSC containing del (5q) as well as somatic driver mutations (ASXL1, JAK2, TET2, TP53, SF3B1). Prior to blast transformation, the pluripotent HSC capable of myeloid and lymphoid differentiation, acquired new KMT2A (MLL) amplification. Absence of MLL amplification in bone marrow suggests this clone seeded into the periphery thereby accounting for its presence in the extramedullary blast transformation but not in the bone marrow. Deletion 17p was present in 2016 accelerated phase ET and appears to have been conserved in the subclone that evolved into myeloid sarcoma. Deletion 20q was first seen in bone marrow after lenalidomide therapy and is present in subclone that evolved into B-cell ALL. Treatment with lenalidomide contributed to the suppression of del(5q) clones in bone marrow but had little effect in extramedullary leukemias ancestral tree in this case report highlights how the relative accessibility of NGS continues to improve our understanding of leukemogenesis, specifically, the predictive significance of large del(5q). It also has the capacity to inform therapeutic choices. Notwithstanding the presence of del (5q), recent studies support that HMA is a superior choice to lenalidomide under the clinical scenario described here. HMA combination regimens [15] currently being investigated may provide durable responses to patients with TP53 mutations.
In conclusion, the advances in NGS technology have made it possible to generate a deep snapshot of the genetic composition of rapidly evolving tumor cells. The development and application of computational algorithms to harness NGS data for ancestral reconstruction could have a dramatic impact in how we practice medicine. Large scale simulation of similar cases to that described in this report would enhance our ability to predict disease progression. It would inform how a specific collection of molecular mutations would favor one therapeutic approach versus another. We propose exploring the use of computational algorithms to resolve patterns of clonal progression to enhance our understanding of leukemogenesis and the contribution of molecular targets, thereby leading to the development of more specific, tumor-tailored therapeutic interventions.