The Classification of Myeloproliferative Neoplasms: Rationale, Historical Background and Future Perspectives with Focus on Unclassifiable Cases

Simple Summary Myeloproliferative neoplasms (MPNs) are clonal hematological disorders, characterized by increased proliferation of the myeloid lineages in the bone marrow. Since their original recognition by William Damashek in 1951, MPNs have been extensively investigated at a clinical-pathological and molecular level. This prompted a progressive refinement of their classification and diagnostic criteria. Uncertainties nonetheless remain in a small (yet consistent) subset of cases, characterized by unconventional and/or overlapping clinical-pathological features. Such cases (referred to as MPN, unclassifiable [MPN-U]) encompass a broad spectrum of entities, including early phase MPNs, terminal (i.e., fibrotic) MPNs, MPNs associated with inflammatory or neoplastic disorders, and poorly characterized MPNs with clinical-pathological mismatch or atypical molecular features. In this review, we discuss the rationale behind the classification and diagnostic criteria of MPNs, focusing on the still open issues concerning MPN-U. Abstract Myeloproliferative neoplasms (MPNs) are a heterogeneous group of clonal hematopoietic stem cell disorders, characterized by increased proliferation of one or more myeloid lineages in the bone marrow. The classification and diagnostic criteria of MPNs have undergone relevant changes over the years, reflecting the increased awareness on these conditions and a better understanding of their biological and clinical-pathological features. The current World Health Organization (WHO) Classification acknowledges four main sub-groups of MPNs: (i) Chronic Myeloid Leukemia; (ii) classical Philadelphia-negative MPNs (Polycythemia Vera; Essential Thrombocythemia; Primary Myelofibrosis); (iii) non-classical Philadelphia-negative MPNs (Chronic Neutrophilic Leukemia; Chronic Eosinophilic Leukemia); and (iv) MPNs, unclassifiable (MPN-U). The latter are currently defined as MPNs with clinical-pathological findings not fulfilling the diagnostic criteria for any other entity. The MPN-U spectrum traditionally encompasses early phase MPNs, terminal (i.e., advanced fibrotic) MPNs, and cases associated with inflammatory or neoplastic disorders that obscure the clinical-histological picture. Several lines of evidence and clinical practice suggest the existence of additional myeloid neoplasms that may expand the spectrum of MPN-U. To gain insight into such disorders, this review addresses the history of MPN classification, the evolution of their diagnostic criteria and the complex clinical-pathological and biological features of MPN-U.


Introduction
The Myeloproliferative Neoplasms (MPNs) are a heterogeneous group of clonal hematopoietic stem cell disorders, characterized by increased proliferation of the myeloid lineages in the bone marrow (BM) [1]. These disorders primarily affect adult to elderly patients with a cumulative annual incidence of 1-5 cases/100,000 [2]. Males are more commonly affected than females, yet relevant differences exist in the epidemiology of each entity. The clinical-laboratory features of MPNs depend on the main lineage of differentiation, the molecular landscape, and the disease stage. The life expectancy of untreated patients is reduced compared to the general population, given the high risk of thrombotic and hemorrhagic complications, BM failure, end-organ damage and blast transformation upon disease progression [3]. The MPNs are traditionally classified into four sub-groups, which include (i) Chronic Myeloid Leukemia (CML); (ii) classical Philadelphia-negative MPNs (Polycythemia Vera, PV; Essential Thrombocythemia, ET; Primary Myelofibrosis, PMF); (iii) non-classical Philadelphia-negative MPNs (Chronic Neutrophilic Leukemia, CNL; Chronic Eosinophilic Leukemia, CEL); and (iv) MPN, unclassifiable (MPN-U) [1]. This classification stems from the decades-long cooperation of hematologists, pathologists, and molecular biologists, which aimed at a better understanding of the pathophysiology, diagnostic criteria, and management of such disorders.
Far from being solely academically relevant, the rationale of MPN classification is instrumental to address the open issues on these tumors. It also supports everyday clinical practice, providing theoretical and practical clues for the diagnosis of difficult cases. To this aim, the present review will address the history of MPN classification and the rationale behind its changes over the years. Special attention will be paid to classical Philadelphia-negative MPNs and MPN-U, which are the most heterogeneous and challenging of all MPNs.

History and Rationale of MPN Classification
The history of MPNs dates back to the nineteenth century, when reports of extreme leukocytosis and/or erythrocytosis highlighted a class of hematological disorders, characterized by markedly increased hematopoiesis [4]. In 1951, William Damashek proposed the first classification of such conditions, which he originally named myeloproliferative disorders (MPDs). Based on Damashek's proposal, the MPDs included CML, PV, ET (referred to as megakaryocytic leukemia), PMF (idiopathic or agnogenic myeloid metaplasia) and erythroleukemia (Di Guglielmo syndrome) [5]. This classification was based mainly on clinical and laboratory findings, with limited contribution of BM morphology (Table 1;  panel A).
Over the years, clinical-pathological studies highlighted consistent differences in the biology and prognosis of MPDs and suggested splitting these entities in two subgroups: (i) myeloleukemia syndromes (CML and erythroleukemia), characterized by single lineage hematopoiesis and a high tendency of blast transformation; and (ii) myeloproliferative syndromes (PV, ET, and PMF) with a more indolent clinical course, non-destructive hematopoiesis and higher rates of fibrotic evolution [6]. While based on disputable assumptions, this distinction was essentially correct, as Di Guglielmo syndrome was to be later listed among acute leukemias/myelodysplastic syndromes [7] and CML disclosed molecular and clinical features clearly distinct from all other MPNs [1,8].
Damashek's classification was embraced in 1976 by the WHO Classification of hematological tumors with the notable inclusion of chronic myeloproliferative diseases (CMPDs), unclassified as a further diagnostic category [9] (Table 1; panel A). According to the WHO standards, such a diagnosis had to be rendered in Philadelphia-negative cases lacking clear-cut features of any other definitive entity.
In the 1980s and early 1990s, BM histology gained progressive relevance in the study of CMPDs and prompted histology-based classifications, including the Working Classification of CMPDs [10] and the so-called Hannover Classification [11]. The former identified three groups of entities: (i) typical CMPDs (corresponding to Damashek's classical PV, ET and  (Table 1; panel B). While addressing the complex histologic landscape of CMPDs, the Working Classification was hampered by the high rate (23%) of unclassifiable cases. The Hannover Classification partially overcame these limitations by adopting a simplified classification scheme and introducing a new clinical-pathological entity (chronic megakaryocytic-granulocytic myelosis) that corresponds to early phase PMF [12] ( Table 1; panel B). Nonetheless, uncertainties remained regarding CNL and CEL (still considered within the spectrum of CML) and on the actual existence of primary (i.e., idiopathic) cases of Myelofibrosis (MF). These issues were addressed by clinical and molecular studies on chronic leukemias [13,14] and by seminal works on PMF from the Cologne group [12]. The 2001 WHO Classification of hematopoietic tumors endorsed these studies and provided an expanded list of CMPDs, including CEL, CNL and Chronic Idiopathic (i.e., Primary) MF as separate entities [15] (Table 1; panel C). This classification relied on the integration of clinical, histological and genetic data, much in the way of the REAL Classification for lymphoid tumors that had been proposed a few years before [15,16]. While not universally accepted [17], this clinical-pathological approach marked a turning point in the history of MPNs.
Following the 2001 WHO Classification, the molecular bases of myeloid tumors were deeply investigated with the discovery of JAK2 and MPL mutations in Philadelphianegative CMPDs [18], KIT mutations in Mastocytosis [19], and PDGFRA, PDGFRB, and FGFR1 rearrangements in subsets of myeloid/lymphoid neoplasms with eosinophilia [20]. All of this was included in the 2008 WHO Classification (Table 1; panel C), which definitively recognized the neoplastic nature of such conditions and abandoned the term CMPD in favor of MPN. Mastocytoses were also listed within the spectrum of MPNs, while myeloid/lymphoid neoplasms with PDGFRA, PDGFRB, and FGFR1 rearrangements were considered separately [21,22]. The 2016 revision of the WHO Classification maintained this scheme with the notable exception of Mastocytoses, which were moved to a separate section due their unique clinical-pathological features [1,23] (Table 1;

Evolution of the Diagnostic Criteria for MPNs
In the early 1970s, the Polycythemia Vera Study Group (PVSG) proposed the first diagnostic criteria for MPNs [24,25]. Being designed for clinical trials, the PVSG consensus had high specificity, but limited sensitivity [4,26]. This prompted the elaboration of more sensitive, integrated clinical-pathological criteria, including the Rotterdam [26] and the 2001 WHO criteria for MPNs [15]. The 2008 and 2016 WHO Classifications endorsed this approach, also considering the molecular and pathogenic features of each entity [1,21].

Evolution of the Diagnostic Criteria for PV
The PVSG diagnostic criteria for PV relied primarily on clinical and laboratory findings and adopted red cell mass (RCM) as the sole indicator of erythroid load. BM evaluation had little (if any) role in the diagnostic workup [4,24]. While still backed by some authors [17], the PVSG criteria were hampered by (i) limited sensitivity; (ii) the need for nuclear medicine to estimate RCM; and (iii) the poor reproducibility of RCM assessment. This was addressed by the Rotterdam [26] and 2001 WHO criteria, which added hemoglobin (Hb) as an indicator of erythroid mass and included BM histology to improve diagnostic accuracy ( Table 2) [15]. The discovery of JAK2 mutations prompted major changes in the diagnosis of PV, as it improved the distinction from reactive erythrocytosis and allowed the detection of early stage disease. Moving from these assumptions, the 2008 WHO Classification listed JAK2 mutations among the major diagnostic criteria for PV. It also lowered Hb levels and added hematocrit as a further surrogate of RCM. As JAK2 mutations are not unique to PV, minor (histological and laboratory) criteria were listed to increase diagnostic specificity [22]. Despite these changes, the 2008 WHO criteria still missed subsets of early phase PV, characterized by overt thrombocytosis and mild erythrocytosis [27]. To detect these "masked PV" cases, the 2016 WHO Classification further lowered the diagnostic thresholds of Hb and hematocrit. The decrease in specificity was compensated with the promotion of BM histology to a major diagnostic criterion (Table 2) [28].

Evolution of the Diagnostic Criteria for ET
Consensus diagnostic criteria for ET were proposed by the PVSG in 1986 [25]. They were designed mainly to exclude alternative causes of thrombocytosis and gave little relevance to BM evaluation (Table 3). Over time, compelling evidence demonstrated the heterogeneity of MPNs with thrombocytosis and highlighted the existence of PV and early PMF cases clinically mimicking ET. The PVSG criteria inadequately distinguished such entities, with detrimental consequences for patient management [12,[29][30][31]. All of this prompted the development of new diagnostic standards (the Rotterdam and 2001 WHO criteria), based on the integration of clinical and pathological data [15].
The 2008 WHO Classification maintained this approach and integrated clinicalpathological and molecular data to increase diagnostic sensitivity, lower platelet thresholds and simplify exclusion criteria [21]. The 2016 WHO Classification made minimal changes to this diagnostic workup, just adding CALR mutations to the molecular derangements typical of ET (Table 3) [1].

Evolution of the Diagnostic Criteria for PMF
The diagnostic criteria for PMF have undergone relevant changes over the years. Prior to the Cologne consensus of 1996, a diagnosis of MF was made only upon documentation of (i) severe BM fibrosis; (ii) anemia; (iii) poikilocytosis; (iv) leukoerythroblastosis; and (v) massive splenomegaly [12]. The description of early phase PMF prompted broader diagnostic standards, considering both pre-fibrotic and full-blown cases. This novel approach was first adopted by the Cologne criteria for Idiopathic MF, which have inspired all subsequent developments [32]. The 2001 WHO Classification endorsed the distinction between pre-fibrotic and overt PMF, but did not provide clear-cut diagnostic criteria for such entities [15]. These limits were addressed by the 2008 WHO Classification, which proposed major and minor criteria based on clinical-pathological and molecular parameters. Early phase and overt disease, however, were lumped together and phase-specific diagnostic parameters were lacking [21]. Separate criteria were eventually proposed by the 2016 WHO Classification, which also added CALR mutations to the list of PMF-associated molecular changes (Table 4) [1]. Of note, the 2008 and 2016 WHO Classifications adopted distinct diagnostic criteria for secondary forms of MF (i.e., post-ET and post-PV MF).

Myeloproliferative Neoplasms, Unclassifiable (MPN-U)
Approximately 5-10% of all MPNs display some clinical, morphological, or molecular features of MPN, but do not fulfill the aforementioned diagnostic criteria or present ambiguous features. These cases (defined as MPN-U) are traditionally clustered into three clinical-pathological groups: (i) early phase MPNs; (ii) advanced fibrotic phase MPNs; and (iii) MPNs with concurrent inflammatory or neoplastic disorders obscuring the clinicalhistological picture [1,33].
In early phase MPN-U, clinical and/or morphological features of a specific entity are not fully developed. With widespread application of molecular screening, pathologists are increasingly confronted with such cases, which account for the vast majority of MPN-Us. Moreover, roughly 50% of MPNs with splanchnic vein thrombosis (SVT) have a clinical phenotype not diagnostic of a specific entity. Until definite criteria are met, such cases are better diagnosed as MPN-U. In some instances, the first diagnosis of MPN is performed in the fibrotic phase and stromal alterations (collagen deposition, increased micro-vessel density, sinusoids ectasia, and bone remodeling) are so advanced that the differential diagnosis between PMF and secondary MF is not feasible [33]. The clinical features of MPN-U at presentation are variable: increased blood cell counts (thrombocytosis, leukocytosis and/or polyglobulia) without organomegaly are more frequent in early phase MPN-U, while cytopenia (thrombocytopenia, leukopenia, and anemia) with splenomegaly and hepatomegaly are usually associated with advanced disease stages. The diagnosis can be challenging and requires the exclusion of mimicking conditions, such as infections and toxin or drug exposure (growth factors, cytokines, or immunosuppressive drugs). In this context, identifying mutations in MPN driver genes (JAK2, CALR, or MPL) or other myeloid neoplasm-associated genes (ASXL1, EZH2, TET2, IDH1, IDH2, SRSF2, and SF3B1) hints to the clonality of hematopoiesis and allows distinction from reactive conditions [1,34]. The same holds true for cytogenetic abnormalities, which can be found in approximately 30% of cases. It is worth noting that a diagnosis of MPN-U cannot be made in cases with genetic lesions defining other myeloid neoplasms (BCR-ABL1 fusion, rearrangements of PDGFRA, PDGFRB, FGFR1 and/or PCM1-JAK2 fusion), if clinical data are incomplete or if biopsy samples are inadequate [1,33].
The evolution and prognosis of MPN-U is variable. In some cases, follow-up allows re-classification within a specific MPN and the prognosis is that of such disease. Patients with MPN-U in the early phases tend to have a relatively favorable prognosis, akin to that of ET or early phase PMF. In these cases, thrombotic-hemorrhagic events are the most frequent complications. In advanced and fibrotic stages, the prognosis is instead poor with possible progression to accelerated or blast phase [33,35].
Literature studies and our own data confirm the heterogeneity of the MPN-U spectrum. As previously highlighted, cases presenting with SVT constitute a unique subtype of MPN with misleading morphology and clinical features. In a study on 29 patients of this type, we found 11/29 (37.9%) cases morphologically consistent with PV, 11 Of the remaining 11/42 (26.2%) cases, 7/42 (16.7%) displayed PMF-like morphology but lacked any of the minor diagnostic criteria, while 4/42 (9.5%) were morphologically consistent with PV, but failed to meet the required Hb threshold values (3 cases) or lacked JAK2 mutations [35]. Overall, these results indicate that refining the diagnostic criteria for conventional MPNs significantly reduces the rate on unclassifiable cases. A consistent proportion of MPNs nonetheless escapes any attempt of classification and may represent a persistence of poorly characterized subsets of myeloid neoplasms. These conclusions are in keeping with a recent study from Deschamps et al. [37], which described the clinical, pathological, and molecular features of a series of 82 MPN-Us. In this cohort, the median age at diagnosis was 49 years (range: 13-79), with a slight female predominance ( , while 20/82 (24.4%) cases were TN; a subset of patients underwent further investigation, which allowed the identification of variants affecting TET2, ASXL1, SRSF2, and RUNX1. Interestingly, reticulin fibrosis was mild (MF-1) to absent (MF-0) in the majority of cases. In keeping with the updated 2016 WHO diagnostic criteria, this suggests that most new MPN-Us are diagnosed in the early, pre-fibrotic stage.

Challenges and Perspectives in the Classification of MPN-U
When dealing with a patient with suspected Philadelphia-negative MPN, several scenarios can be recognized (Figure 1). In many instances, the clinical, histological, and molecular findings allow a definite diagnosis [38,39]. In rare cases, however, clinical-pathological and/or molecular mismatch contrasts with WHO-defined entities and prevents definite conclusions. Such cases may expand the spectrum of acknowledged MPN-Us, pinpointing to future developments on MPN classification. The following paragraphs will address these unconventional cases, presenting real-life examples and discussing a general framework for their interpretation.

MPN with Clinical-Morphological Mismatch
In the proper clinical setting, histology provides invaluable information to make a definite diagnosis of MPN. In particular, megakaryocyte morphology and hypercellular hematopoiesis represent cardinal diagnostic traits, the latter being minor-to-absent in ET and in advanced phases of fibrotic MPN. Particularly in early phase MPNs, the lack of fully-developed morphologic features may conceal a diagnosis of clonal disease, or its punctual classification. Furthermore, morphologic anomalies of the hematopoietic lineages and/or increased monocyte, eosinophil, and basophil counts may challenge a diagnosis of MPN, leaning towards myelodysplastic syndrome (MDS) or MDS/MPN conditions.

Cases with MPN-like Clinical Findings, Lacking MPN Morphologic Criteria
As previously reported, morphologic patterns on BM biopsy constitute a major and necessary criterion to make a diagnosis of MPN (Tables 2-4), despite that histologic evaluation is hampered by some degree of interobserver variability. To this regard, the differential diagnosis between early PMF and ET constitutes critical and long-debated grounds for interpretation [40]. In this setting, a provisional diagnosis of MPN can be provided with the caveat that definite subtyping may require follow up. In rarer instances (which are exemplified hereafter) unconventional clinic-pathologic features may hinder not only the definition of a specific MPN, but also the diagnosis of MPN itself.

Case 1-Synopsis and Discussion
A 41-year old male presents with a 5-year history of erythrocytosis and high hematocrit levels. Laboratory test at presentation disclose high Hb (17.8 g/dL) and Hct (54%), normal WBC (6.7 × 10 9 /L) and platelet counts (203 × 10 9 /L), and erythropoietin (Epo) levels below local laboratory ranges (3.9 mU/mL [4.3-29]). A JAK2 V617F mutation with 2.8% variant allele frequency (VAF) is detected on peripheral blood granulocytes (data on exon 12 and/or additional exon 14 mutations not available). The overall clinical presentation is suspected for PV and prompts a BM evaluation. A BM biopsy ( Figure 2) features cellularity within age-related limits, preserved maturation of erythroid and myeloid lineages, and only a minor increase in megakaryocytes. The latter are sparse in distribution, with predominant mature-type morphology and only rare, scattered large forms. Thus, the lack of clear-cut major criterion 2 ( Table 2) does not allow a diagnosis of PV and warrants a provisional characterization as MPN-U.
In the present case, the histology is not inconsistent with the clinical diagnosis of PV, but it is not robust (particularly, some clustering of megakaryocytes with scattered large forms are disputable). A pre-morphologic diagnosis of MPN (i.e., a diagnosis made without full-blown MPN histology) is not common [41]. Indeed, it may occur in clinical settings where awareness of early phase MPNs suggests comprehensive laboratory and molecular screening, including quantitative PCR for low-burden driver gene mutations [42]. In such a context, it is worth noting that low VAF of JAK2 V617F may be due to multiple base mutations in exon 14, which may hamper primer annealing and affect the quantification of the mutational burden [41]. Thus, searching for additional exon 14 mutations may be justified in cases with inexplicably low burdens of JAK2 V617F . and/or additional exon 14 mutations not available). The overall clinical presentation is suspected for PV and prompts a BM evaluation. A BM biopsy (Figure 2) features cellularity within age-related limits, preserved maturation of erythroid and myeloid lineages, and only a minor increase in megakaryocytes. The latter are sparse in distribution, with predominant mature-type morphology and only rare, scattered large forms. Thus, the lack of clear-cut major criterion 2 ( Table 2) does not allow a diagnosis of PV and warrants a provisional characterization as MPN-U.

Cases with MPN-like Morphology, Lacking MPN Clinical Criteria
Far more common than the previous case is the instance of histopathologic and/or molecular findings fully supportive of MPN, but with clinical features obscuring either the diagnosis or the subtyping of the clonal process.

Case 2-Synopsis and Discussion
A 32-year old pregnant woman refers to the Emergency Unit for abdominal pain. Sonography reveals SVT of the splenic-portal axis and blood tests disclose anemia (Hb = 10.5 g/dL; Hct = 32%), leukocytosis (WBC = 18.2 × 10 9 /L), and thrombocytosis (Plt = 596 × 10 9 /L) with low Epo levels (3.3 mU/mL [4.3-29]), and JAK2 V617F mutation. A BM biopsy (Figure 3a-c) discloses PV changes with hypercellularity, expansion of the three hematopoietic lineages, clustering of polymorphic megakaryocytes, and no fibrosis. Despite this, the lack of the first major criterion hampers a formal diagnosis of PV.
The present case depicts the prototypic features of "masked PV", presenting with SVT. A further bias is the concurrent pregnancy, which itself favors deep vein thrombosis. Such a clinical event may occur in any subtype of MPN, either at diagnosis or during the course of the disease. A higher thrombotic risk is related to a hyperinflammatory state and neutrophilia, which are more commonly experienced in PV [43]. Besides the diagnostic issues, it should be stressed that thrombotic events pose the highest threat for patients with early phase MPN. Despite the risk factors for vascular events are only partially defined, major determinants include JAK2 status, age, neutrophil count, and a positive history of thrombosis [43][44][45].

Case 3-Synopsis and Discussion
A 44-year old female with unremarkable past clinical history refers to the Hematology Clinic for persistent thrombocytosis (Hb = 15.2 g/dL; Hct = 44.8%; WBC = 12.65 × 10 9 /L [N = 9.58 × 10 9 /L]; Plt = 742 × 10 9 /L). Epo levels are close to the lower limit of normal (4.29 mU/mL; range: 4.3-29), LDH is not increased (206 U/L, range: 135-214) and JAK2 V617F mutation is detected (VAF = 5.5%). Splenomegaly is not documented at clinical examination and sonography. BM biopsy (Figure 3d-f) discloses a hypercellular marrow with features of panmyelosis, maturing erythroid and myeloid lineages and polymorphous, and loosely clustering megakaryocytes with frequent giant hyperlobated forms. A minor increase in reticulin fibers is observed, consistent with MF-1 according to WHO criteria. This case displays borderline features between a pre-polycythemic phase of PV (highly supported by the very typical morphologic pattern) and pre-fibrotic PMF. hyperlobulated ("staghorn") cells. No increase in reticulin fibers and CD34+ blasts is observed; the lympho-plasmacytic infiltrate is unremarkable.
The global picture strongly favors a diagnosis of MPN, but the differential diagnosis between ET and PV cannot be definitively settled. The former is favored by thrombocytosis and by prominent ET-type megakaryocyte morphology, and the latter is supported by the coexistence of neutrophilia alongside with normal Hb levels (possible effect of the concurrent thalassemic trait) [46]. A diagnosis of ET is ruled out by BM histology. In contrast to case 1, the still low JAK2 V617F allele burden could be responsible for the blood count, which is not fully consistent with PV. In similar cases, concurrent hematopoiesis-modifying factors (e.g., thalassemic trait, vitamin deficiencies, malabsorption, etc.) should be investigated. In addition, a minor increase in fibrosis should warrant closer monitoring.
The global picture strongly favors a diagnosis of MPN, but the differential diagnosis between ET and PV cannot be definitively settled. The former is favored by thrombocytosis and by prominent ET-type megakaryocyte morphology, and the latter is supported by the coexistence of neutrophilia alongside with normal Hb levels (possible effect of the concurrent thalassemic trait) [46].

MPN with Unusual Morphology
In some cases, very unusual morphologic features or non-typical clinic-laboratory findings may challenge the diagnosis of MPN itself or suggest the differential diagnosis with other myeloid neoplasms (e.g., MDS/MPN).

Case 5-Synopsis and Discussion
A 32-year old male with unremarkable past clinical history presents with persistent thrombocytosis (Hb = 15.6 g/dL; Hct = 46.1%; WBC = 6.76 × 10 9 /L; Plt = 745 × 10 9 /L; Epo and LDH levels within normal ranges). JAK2 V617F mutation is detected (VAF = 10.1%) on peripheral blood granulocytes. A BM biopsy (Figure 4a-c) features striking hypocellularity without morphologic atypia. A slight decrease in the myeloid-to-erythroid ratio is noted with mildly increased, polymorphic megakaryocytes consisting of mature-looking to giant, hyperlobulated forms. There is no evidence of BM fibrosis and CD34+ blasts are not increased. The clinicopathologic synopsis supports MPN, but the very unusual morphologic picture favors a provisional diagnosis of MPN-U. Hypocellularity is uncommon in clonal myeloid disorders in general and barely reported in both early and chronic phase MPN. It may nonetheless be encountered in late-stage disease with the classical picture of MF with myeloid metaplasia. In this setting, the BM niche is overwhelmed by the stromal component with consequent peripheralization of hematopoietic cells. In the current case, the diagnosis of MPN is strongly supported by the laboratory findings, by the presence of a driver mutation, and by megakaryocyte morphology. Even though most clues point towards ET, the relative increase in the erythroid progenitors and Hb levels at the upper limit of normal cannot definitively rule out early phase PV with ET-like presentation and very atypical BM morphology.

Case 6-Synopsis and Discussion
An 80-year old male presents with longstanding monocytosis and progressive thrombocytosis without anemia (Hb = 13.4 g/dL; Hct = 41.3%; WBC 5.9 × 10 9 /L [monocytes 1.42 × 10 9 /L, 24%]; Plt = 1.348 × 10 9 /L). Epo levels are within normal limits, while LDH levels are increased (290 U/L; range: 135-225). Morphologic assessment of a BM smear is not informative due to absent cellularity, whereas a BM biopsy (Figure 4d-f) features hypercellularity with increased myeloid-to-erythroid ratio, myeloid left shifting, and a well-represented CD14+ monocyte component. The megakaryocytes display clustering with frequent giant hyperlobated forms. Small megakaryocytes with maturation defects are also present. Reticulin stain and blast count are unremarkable. Targeted next generation sequencing (NGS) on the peripheral blood cells documents mutations of TET2 (VAF = 37%), JAK2 (VAF = 26%) and CALR (VAF = 28%); the karyotype is normal.The clinic-pathologic picture is consistent with a myeloid neoplasm with monocytosis and poses the differential diagnosis between PMF and chronic myelomonocytic leukemia. While the overall histological picture is more in keeping with MPN, the left shifting of the myeloid series and the maturation defects of the megakaryocytes may, in contrast, suggest multi-lineage dysplasia. However, the mutational profile favors a JAK2/CALR-driven MPN. TET2 mutations are of little value, as they can be present (likely as early clonal events) both in MPN, MDS/MPN, and in clonal hematopoiesis of indeterminate potential (CHIP) [47][48][49][50][51]. As to the unusual coexistence of driver gene mutations, a JAK2/CALR double-mutated genotype has been reported rarely in MPNs and does not correlate with specific clinical phenotypes [49,52]. Finally, it should be stressed that monocytosis in MPNs is mostly observed alongside fibrotic progression and may be assumed as an indicator of accelerated phase disease [53,54]. This was not the case for our patient.

MPN with Unconventional Molecular Features
Molecular genetics is a cardinal tool in the workup of MPNs, as it contributes to the diagnosis, classification, and prognostic stratification of these disorders. The identification of myeloid neoplasm-related gene derangements constitutes a major diagnostic criterion (i) to prove the clonal nature of the disease (particularly in TN cases); and/or (ii) to support/exclude specific disease subtypes [55]. With its increasing availability in the clinical setting, targeted NGS also allows the detection of atypical mutations in driver genes and of genetic derangements impacting on disease progression [56][57][58]. The quantitative assessment of VAF can also provide clinically relevant information on the disease burden and on the risk of adverse events. Indeed, the VAF of driver genes increases alongside myelofibrotic evolution, and high JAK2 V617F burdens correlate with the risk of thrombosis [45,48,59,60]. Together with mutational analyses, karyotyping proves informative to assess the risk of progression, particularly in the myelofibrotic setting [61].

JAK2-Negative Erythrocytosis
A definite diagnosis of PV should be rendered with great caution in the absence of identifiable JAK2 mutations, as these occur in >95% of cases [1,62]. A common instance related to this scenario is the occurrence of cases with low-burden driver gene mutations (i.e., VAF falling below the sensitivity of the available techniques) [41]. Less frequently, a clinical and morphologic picture fully consistent with PV may be sustained by a JAK2 wt clone. This imposes a challenging assessment of alternative congenital or acquired states of erythrocytosis [62]. Interestingly, sporadic JAK2 wt MPN cases featuring the clinical phenotype of PV have been associated with mutations of CALR, either canonical (type 1) [63] or noncanonical [64]. Rare JAK2 wt cases, some of which with PV features, may also carry mutations in LNK/SH2B3 [65]. This genetic event has been associated with an increased risk of blastic transformation in MPN [66], as well as with occasional cases of idiopatic erythrocytosis [62].

MPN with "High Risk" Molecular Features
While molecular testing is pivotal to guide MPN diagnosis [67], in rare instances it may yield puzzling results, which need to be interpreted after integration with clinical, laboratory, and histological data. This is specifically the case of genes rarely mutated in MPNs and associated with an aggressive clinical course.
As to the unusual coexistence of driver gene mutations, a JAK2/CALR double-mutated genotype has been reported rarely in MPNs and does not correlate with specific clinical phenotypes [49,52]. Finally, it should be stressed that monocytosis in MPNs is mostly observed alongside fibrotic progression and may be assumed as an indicator of accelerated phase disease [53,54]. This was not the case for our patient.

MPN with Unconventional Molecular Features
Molecular genetics is a cardinal tool in the workup of MPNs, as it contributes to the diagnosis, classification, and prognostic stratification of these disorders. The identification of myeloid neoplasm-related gene derangements constitutes a major diagnostic criterion (i) to prove the clonal nature of the disease (particularly in TN cases); and/or (ii) to support/exclude specific disease subtypes [55]. With its increasing availability in the clinical setting, targeted NGS also allows the detection of atypical mutations in driver genes and of genetic derangements impacting on disease progression [56][57][58]. The quantitative assessment of VAF can also provide clinically relevant information on the disease burden and on the risk of adverse events. Indeed, the VAF of driver genes increases alongside myelofibrotic evolution, and high JAK2 V617F burdens correlate with the risk of thrombosis [45,48,59,60]. Together with mutational analyses, karyotyping proves informative to assess the risk of progression, particularly in the myelofibrotic setting [61].

Case 7-Synopsis and Discussion
A 46-year-old woman presents with a long-lasting history of thrombocytosis (mean platelets: 500 × 10 9 /L) and a putative clinical diagnosis of ET treated with low-dose aspirin. A BM biopsy ( Figure 5) is performed 20 years after clinical onset and shows hypercellular, trilinear hematopoiesis with normal myeloid-to-erythroid ratio, mildly left shifted granulopoiesis, and increased, mature-looking megakaryocytes. Blasts account for 3-4% of the hematopoietic cells and reticulin fibrosis is not increased. Driver mutations of JAK2, CALR and MPL are not detected, but targeted NGS documents the L56S RUNX1 variant. The overall morphological and molecular findings exclude ET and suggest an alternative diagnosis of RUNX1-mutated myeloid neoplasm with thrombocytosis.
Germline RUNX1 mutations are reported in inherited forms of platelet disorders with a predisposition to myeloid malignancies [68]. Acquired RUNX1 derangements are instead relatively common in MDS, MDS/MPN, AML, and in contexts akin to CHIP [69]. In the present case, the unremarkable family history, the clinical presentation, and the histological findings exclude such diagnoses and favors MPN-U. Of note, rare cases of MPN feature RUNX1 mutations and typically undergo myelofibrotic and/or blastic progression [58,70]. This and 3 similar cases published by Cattaneo and colleagues [55] disclose the unique features of an MPN with "high risk" genetics and an indolent, ET-like clinical course. instead relatively common in MDS, MDS/MPN, AML, and in contexts akin to CHIP [69]. In the present case, the unremarkable family history, the clinical presentation, and the histological findings exclude such diagnoses and favors MPN-U. Of note, rare cases of MPN feature RUNX1 mutations and typically undergo myelofibrotic and/or blastic progression [58,70]. This and 3 similar cases published by Cattaneo and colleagues [55] disclose the unique features of an MPN with "high risk" genetics and an indolent, ET-like clinical course.
From a clinical perspective, an overlap between CHIP and early TN MPN is sustained by the indolent clinical course and by the slightly increased thrombotic risk of both conditions [48,73,74]. In line with this, by applying highly sensitive molecular techniques, we could document CHIP-related changes in >80% of TN MPNs with ET-like presentation and chronic-indolent follow up [55]. This data points toward a close relationship between CHIP and some early phase TN MPNs. They also pose intriguing questions on the boundaries between such entities.
It should also be noted that large cohort studies on TN MPNs [56] and CHIP [75,76] report high allele burdens for the above-mentioned variants in a substantial fraction of patients, implying a relevant role for such (even "high risk") imbalances in sustaining clonal hematopoiesis.
Overall, subsets of CHIP may thus represent precursor hematopoietic clones, characterized by normal blood cell counts and intrinsic potential to TN MPN evolution [55,72]. Further clinical-pathological and molecular studies are needed to test this fascinating possibility.

Conclusions
Philadelphia-negative MPNs are a heterogeneous group of hematopoietic disorders, characterized by distinct clinical and biological features. Over the years, the classification and diagnostic criteria of such entities have undergone major changes and have been progressively refined. Despite this, a small (yet consistent) subset of cases lacks a precise definition. Such MPN-Us represent either early or advanced disease stages, as well as MPNs with hardly classifiable biological features. A precise characterization of these entities is made possible by the thorough integration of clinical, histological, and molecular studies. This approach will hopefully contribute to a better understanding of MPN biology and to further improvements in the clinical managements of patients.