Myeloproliferative neoplasms (MPNs) are a group of myeloid neoplasms characterized by bone marrow hyperplasia and overproduction of at least one lineage of blood cells. The current subclassification of MPNs is based on changes in blood cell counts, and hematopoietic lineages in the bone marrow that display hyperplasia and dysplasia. Primary myelofibrosis (PMF) is a subtype of BCR::ABL1-negative classic MPN, which also includes polycythemia vera (PV) and essential thrombocythemia (ET). The proliferation of abnormal megakaryocytes and varying degrees of fibrosis are defining features of PMF. PMF also typically presents with splenomegaly due to granulocytic proliferation and extramedullary hematopoiesis, and many patients show constitutional symptoms of a hypermetabolic state due to changes in inflammatory cytokines. Recent updates of the 5th edition of the World Health Organization (WHO) Classification of Hematolymphoid Tumors (WHO-HAEM5) [1] and the International Consensus Classification (ICC) [2] have further refined PMF into early, prefibrotic, and overt fibrotic stages. Secondary myelofibrosis (SMF) can present in the later stages of other myeloid neoplasms, particularly other MPNs (post-ET and post-PV MF) and myelodysplastic/myeloproliferative neoplasms (MDS/MPN). It is necessary to differentiate between ET and PV with mild MF and prefibrotic PMF [1]. However, post-PV and post-ET SMF ^[3][4][3,4] can be indistinguishable from PMF when no clear clinical history of PV or ET is documented in patients presenting with myelofibrosis (Figure 1). PMF and SMF are frequently studied together and are clinically managed similarly. Bone marrow fibrosis can also occur in reactive conditions, such as infections, autoimmune disorders, and other malignancies. In the published literature, the term MF is usually reserved for bone marrow fibrosis related to myeloid neoplasms; bone marrow fibrosis is a general term used for other secondary fibrosis [5].

MF is a distinctive entity among MPNs, signified by a higher risk of transformation to acute myeloid leukemia (AML). Disease progression of MF can also present as refractory cytopenia, progressive leukocytosis, or refractory progression with an increasing fibrotic burden [6]. With the availability of molecular testing, especially next-generation sequencing (NGS) in clinical laboratories, mutational profiling has transformed the diagnostic and classification paradigms for myeloid neoplasms. Detecting the genetic alterations of MF is not only required for diagnosis as clonal evidence but also provides crucial information to help understand its pathobiology in relation to other myeloid neoplasms. Reflecting the expanding utilization of molecular testing and NGS in clinical laboratories, a bibliometric analysis of publications on MPN from 2001 to 2022 indicated that “gene mutations” has been the top keyword for published studies over the past two decades [7]. However, several aspects of MF remain poorly understood, including the biologic and molecular basis of fibrosis as a distinct feature of PMF, potential biologic distinctions between PMF and post-PV or post-ET MF, and differences between proliferative and dysplastic/cytopenic forms of MF.

The discovery of recurrent mutations in Janus kinase 2 (JAK2), calreticulin (CALR), and myeloproliferative leukemia oncogene (MPL) as driver mutations has transformed the diagnostic approach of MPN, as evident in the revisions of WHO classifications ^[1][8][9][1,8,9]. Clonal evidence, supported by the presence of a driver or other mutations commonly associated with various myeloid neoplasms, is crucial for definitive diagnosis. Both JAK2 and MPL encode proteins that activate the JAK/STAT signaling pathway, which is essential for signal transduction from erythropoietin (EPO), thrombopoietin (TPO), and granulocyte colony-stimulating factor (G-CSF) receptors. The pathobiology and diagnostic relevance of activating mutations in JAK2, CALR, and MPL have been extensively investigated in the clinical setting. JAK2 V617F mutation is associated with an increased risk of thrombosis, and a high allele burden is associated with disease progression [10]. MPL encodes the TPO receptor, and mutations, usually at codon W515, lead to constitutively active signaling independent of ligand binding. The interaction between MPL and altered calreticulin encoded by mutant CALR results in MPL hyperactivity [11]. CALR and MPL mutations are typically exclusive to ET and PMF and very rarely occur in PV [10]; however, JAK2 V617F mutation remains the most common driver mutation in PMF, reported in 50–60% of cases, followed by CALR mutations in 25–35% and MPL mutations in 5–10% cases ^[12][13][12,13]. Interestingly, JAK2 exon 12 mutations [14], which are also activating mutations, have not been documented in ET or PMF. All oncogenic CALR mutations are frame-shifting insertions or deletions (indels) that alter the C-terminal end of calreticulin from negatively charged acidic amino acids, aspartic acid (D)- and glutamic acid (E)-rich, to positively charged basic amino acids, arginine (R)- and lysine (K)-rich, removing the endoplasmic reticulin retention signal KDEL. Mutant calreticulin can be secreted and functions as a cytokine, retaining its ability to bind to MPL in the CALR-mutated clone [15]. In PMF, type 1 CALR mutations (51 bp deletion, L367Tfs*46) are approximately three times more prevalent than type 2 (5 bp insertion, K385Nfs*47) [13], with phenotypic variations observed among CALR mutation types [16]. In PMF, type 1 CALR mutations correlate with lower leukocytosis, lower bone marrow cellularity, and an increased number of megakaryocytes [13], while type 2 mutations align more closely with the phenotype of cases harboring JAK2 V617F [17]. In the vast majority of MPN cases, driver mutations in JAK2, CALR, and MPL are mutually exclusive. However, there have been occasional reports of cases exhibiting coexistence of JAK2 V617F, MPL, and/or CALR mutations ^[18][19][18,19]. Such cases likely involve distinct subclones of neoplastic cells harboring different driver mutations, as demonstrated by a single-cell sequencing study [20], although instances of dual mutations in a single clone have also been documented [21]. Approximately 10% of MPN cases lack detectable canonical mutations in JAK2, CALR, or MPL, categorizing them as triple-negative (TN) MPNs. A small subset of these cases may not truly be TN, as other rare gain-of-function mutations in one of these three genes, particularly MPL, have been reported ^{[22][23][24][25][26]}[22,23,24,25,26]. True TN cases often harbor mutations outside of these three genes, confirming clonal hematopoiesis. However, these mutations, which are also prevalent in other myeloid neoplasms, are not considered driver mutations of MPNs. Despite the availability of NGS tests for clinical analysis, the driver mutations of TN cases have not yet been determined, even with comprehensive whole-exome sequencing (WES) studies. One candidate driver, SH2B3 mutation, has been identified in a subset of TN MPNs [27]. However, SH2B3 mutations and other driver mutations are not mutually exclusive. The pathogenic drivers of TN MPNs are either heterogeneous non-recurrent mutations, more complicated alterations that evade ready identification by currently available methods, or with mechanisms not yet recognized. Further exploration to understand the regulatory sequences within the non-coding regions of the human genome may shed light on the drivers and molecular pathogenesis of TN MPNs. With the accumulation of mutation profiling data from clinical studies, it is now clear that over 50% patients with MPNs harbor mutations in addition to driver mutations. Among the classic MPNs, PMF has the highest prevalence of additional mutations. With targeted sequencing of myeloid neoplasm-related genes, additional mutations have been reported in approximately 50% of PV and ET cases, and as high as 80% of PMF cases ^[28][29][30][28,29,30]. PMF also harbors a higher number of mutations than PV or ET ^[28][29][31][28,29,31] (Figure 2). Although additional mutations are not considered driver mutations of MPNs, they help establish the clonal nature of TN patients and have been integrated into the major diagnostic criteria of MPNs ^[1][2][1,2]. A query of the American Association for Cancer Research (AACR) Project GENIE public database in cBioportal [32] found 299 samples from 202 cases documented as PMF (https://genie.cbioportal.org/study?id=6562046bb01fff74fbb6c576 (accessed on 25 November 2023)). In these 299 samples, in addition to JAK2 (44.8%), CALR (14.7%), and MPL (9.4%) mutations, the prevalence of other mutations is similar to those reported by other studies ^{[29][31][33][34]}[29,31,33,34]. Table 1 lists the prevalence of relatively frequent non-driver mutations and the most common mutations or mutation types cataloged in the GENIE database. In addition to the mutations detected in sequencing studies, cytogenetic abnormalities have been reported in 30–57% of PMF cases. However, none of the abnormal karyotypes are specific to PMF [35]. The spectrum of additional mutations detected in PMF did not differ from that detected in PV or ET. However, mutations in genes involved in chromosome modification (ASXL1 and EZH2), DNA methylation (DNMT3A), and RNA splicing (SRSF2, ZRSR2, and U2AF2) were more frequently observed in PMF ^[36][37][36,37]. Follow-up studies have shown that most somatic mutations in MPN are present at diagnosis, instead of developing during disease progression ^[38][39][38,39]. The mutation profiles were similar in PMF and SMF. ASXL1 mutation has the highest prevalence, close to 50% in PMF and 30–40% in SMF in some studies ^[40][41][40,41]. Yan et al. studied 258 consecutive PMF patients with 275 samples by sequencing 27 genes, with 17 patients tested on at least two time points, and found that the variant allele frequency (VAF) of ASXL1 mutations was relatively stable during the disease process [42]. Luque et al. reported that ZRSR2 and NFE2 mutations were more common in SMF [41].