Myeloid/lymphoid neoplasms with eosinophilia and tyrosine kinase gene fusions: reevaluation of the defining characteristics in a registry-based cohort

In a registry-based analysis of 135 patients with “myeloid/lymphoid neoplasms with eosinophilia and tyrosine kinase gene fusions” (MLN-TK; FIP1L1::PDGFRA, n = 78; PDGFRB, diverse fusions, n = 26; FGFR1, diverse, n = 9; JAK2, diverse, n = 11; ETV6::ABL1, n = 11), we sought to evaluate the disease-defining characteristics. In 81/135 (60%) evaluable patients, hypereosinophilia (>1.5 × 109/l) was observed in 40/44 (91%) FIP1L1::PDGFRA and 7/7 (100%) ETV6::ABL1 positive patients but only in 13/30 (43%) patients with PDGFRB, FGFR1, and JAK2 fusion genes while 9/30 (30%) patients had no eosinophilia. Monocytosis >1 × 109/l was identified in 27/81 (33%) patients, most frequently in association with hypereosinophilia (23/27, 85%). Overall, a blast phase (BP) was diagnosed in 38/135 (28%) patients (myeloid, 61%; lymphoid, 39%), which was at extramedullary sites in 18 (47%) patients. The comparison between patients with PDGFRA/PDGFRB vs. FGFR1, JAK2, and ETV6::ABL1 fusion genes revealed a similar occurrence of primary BP (17/104, 16% vs. 8/31 26%, p = 0.32), a lower frequency (5/87, 6% vs. 8/23, 35%, p = 0.003) of and a later progression (median 87 vs. 19 months, p = 0.053) into secondary BP, and a better overall survival from diagnosis of BP (17.1 vs. 1.7 years, p < 0.0008). We conclude that hypereosinophilia with or without monocytosis and various phenotypes of BP occur at variable frequencies in MLN-TK.


INTRODUCTION
The recently published World Health Organization (WHO) 2022 classification and the International Consensus Classification of Myeloid and Lymphoid Neoplasms (ICC-MLN) define a distinct subcategory of myeloid neoplasms as "myeloid/lymphoid neoplasms with eosinophilia and tyrosine kinase gene fusions" (MLN-TK) [1,2]. This category name has changed from the previous "myeloid/lymphoid neoplasms with eosinophilia and rearrangement of PDGFRA, PDGFRB or FGFR1, or with PCM1::JAK2" (MLN-eo) [3], to specify the underlying molecular genetic changes and to include cases with ETV6::ABL1, FLT3 fusions or other TK fusion genes. MLN-TK are driven by rearrangements/fusion genes with involvement of PDGFRA, PDGFRB, FGFR1, JAK2, ABL1, or FLT3. This definition implicates eosinophilia as a recurrent finding, which therefore serves as the main trigger for initiation of distinct cytogenetic and molecular analyses conferring to the identification of disease-defining underlying TK fusion genes. Beside eosinophilia, blast phase (BP) in bone marrow (BM) or at extramedullary sites (extramedullary disease, EMD) of myeloid or lymphoid origin, initially often diagnosed as "myelosarcoma" or "high-grade lymphoma", is present at diagnosis (primary BP) or develops during follow-up (secondary BP).
Single case reports and small series described absence of eosinophilia and/or presence of monocytosis in association with distinct TK fusion genes. Due to the absence of eosinophilia, the diagnosis of a TK-fusion driven MLN may therefore be delayed or even completely missed. Within the "German Registry for Disorders of Eosinophils and Mast Cells (GREM)", we sought to evaluate incidence, the incidence of eosinophilia and monocytosis and the phenotype and prognosis of BP within this distinct subcategory of myeloid neoplasms.
In the current analysis, OS was analyzed in the cohort of 135 patients (male 126/135; median age 49 years, range 19-80), in either chronic phase (CP) from time of diagnosis (n = 110, including 13 patients with progression to secondary BP) or in BP from time of diagnosis of BP (n = 38, primary BP, n = 25; secondary BP, n = 13). Of note, the 13 patients with secondary BP are included in both cohorts (Table 2). Data on absolute and relative counts of eosinophils and monocytes at time of diagnosis were available for 81 patients: FIP1L1::PDGFRA (n = 44), PDGFRB (n = 16), FGFR1 (n = 6), JAK2 (PCM1::JAK2, n = 7, BCR::JAK2, n = 1) or ETV6::ABL1 (n = 7, Table 1). All patients gave written informed consent. Data collection was compliant with the Declaration of Helsinki and approved by the ethics committee of the Medical Faculty Mannheim at the University Heidelberg, Germany.

Cytogenetics and molecular analyses
Cytogenetics and fluorescence in situ hybridization (FISH) analyses were performed on BM according to standard procedures. Specific nested reverse transcription polymerase chain reaction (RT-PCR) was performed for confirmation of suspected fusion genes in all patients [9,17,18].

Statistical analyses
All clinical and laboratory parameters including peripheral blood cell counts are expressed as median and range. Overall survival (OS) was determined from date of diagnosis to date of death or last contact and calculated by using the Kaplan-Meier method. Pearson correlation analysis was performed for the correlation between two parameters. Differences in the distribution of continuous variables between categories were analysed by Mann-Whitney test (for comparison of two groups). For categorical variables, Fisher's exact test was used. P < 0.05 (two-sided) were considered significant. All statistical analyses were performed using GraphPad Prism Software, Inc. version 7 and SPSS (version 28.0; IBM-Corporation, Armonk, NY, USA).

DISCUSSION
Common features of the vast majority of TK-fusion driven myeloid neoplasms include an underlying chronic myeloid neoplasm with a high incidence of concurrent primary BP or progression to secondary BP. BP can be myeloid or lymphoid and is identified in the BM or at extramedullary sites (EMD). In the EMD, initial diagnosis frequently states "myelosarcoma" or "T-cell lymphoma", while in the BM, the differentiation between a de novo myeloid/ lymphoid/biphenotypic acute leukemia and a myeloid or lymphoid BP also remains challenging. We have reported on several patients with suspected primary lymphoma or de novo acute leukemia in which the underlying TK fusion gene was only identified because of poor response to intensive chemotherapy or even allogeneic SCT and persisting eosinophilia [7]. In the currently reported cohort of 135 MLN-TK patients, incidence, phenotype and prognosis of BP was highly variable within the various cohorts of MLN-TK. BP occurred equally distributed either in the BM or as EMD in approximately 30% of patients. It was primary in approximately 70% of patients with a lower relative frequency of 16% in patients with PDGFRA/PDGFRB fusion genes as compared to 26% in patients with FGFR1, JAK2 and ETV6::ABL1 fusion genes. In patients with PDGFRA/PDGFRB fusion genes, secondary BP only occurred in 6% of patients after a median of 87 months because >90% of patients achieved durable complete hematologic, complete cytogenetic (PDGFRB) and complete molecular (FIP1L1::PDGFRA) remissions on imatinib.
In contrast, neither ponatinib on FGFR1 [20], ruxolitinib on JAK2 [10,16] nor imatinib/nilotinib/dasatinib on ETV6:ABL1 fusions [16] have shown a similar efficacy than imatinib on PDGFRA/PDGFRB fusions ( Fig. 1A-C). Of interest, the FIGHT-203 study presented promising results on pemigatinib in patients with FGFR1 fusions in CP and to a lesser extent in BP [21]. In a recent literature review of a heterogenous cohort of 66 PCM1::JAK2 positive patients, Kaplan et al. reported on 11 ruxolitinib-treated patients [12]. However, the authors did not draw conclusions on its effect on survival because of the small cohort and because analysis on survival was complicated by the fact that 5 of these patients received a subsequent allogeneic SCT with a 5-year survival of 75% [12]. In consequence, patients with FGFR1, JAK2, and ETV6::ABL1 fusion genes progressed more often (35%) and faster (median 19 months) into secondary BP than patients with PDGFRA/PDGFRB fusion genes. The inferior prognosis of patients with FGFR1, JAK2, and ETV6::ABL1 fusion genes with a median 5-year survival of approximately 50-60% is therefore related to the more aggressive phenotype and the lack of effective and durable conventional treatment [16,[22][23][24][25][26]. In line with recently published data on patients with FGFR1 [25] and JAK2 fusion genes [12], data confirm that the poor prognosis of primary and secondary BP can only be overcome by allogeneic SCT (Table 4 and Fig. 1A-D).
Consistent hypereosinophilia >1.5 × 10 9 /l in more than 90% of patients was only observed in association with FIP1L1::PDGFRA and ETV6::ABL1 fusion genes, although we acknowledge an obvious Table 4. Outcome after allogeneic stem cell transplantation according to fusion gene and disease phase (n = 25; CP, n = 12; BP, n = 13). ascertainment bias in that only cases with eosinophilia are routinely screened for distinct TK fusion genes, particularly FIP1L1::PDGFRA. In patients with PDGFRB fusion genes, hypereosinophilia >1.5 × 10 9 /l was present in only 50% of patients and even absent (≤0.5 × 10 9 /l) in 25% of patients. Lack of eosinophilia was also evident in patients with FGFR1 fusions (Table 5). These findings are in line with literature reports on the impact of the FGFR1 partner gene on phenotype [27]. While ZMYM2::FGFR1 positive patients frequently present with the combination of a T-cell lymphoma/T-ALL/mixed phenotype acute leukemia and eosinophilia, BCR::FGFR1 positive patients usually present with a MPN/CML-like phenotype but a much lower incidence of hypereosinophilia >1.5 × 10 9 /l, in the literature overall only reported in 3/21 evaluable patients [11,. Diagnosis of a TK fusion gene driven MLN may be missed due to the lack of eosinophilia and subsequent diverse morphological diagnoses [5]. A significant proportion of patients were not initially diagnosed as MLN-TK or chronic eosinophilic leukemia but rather as subtype of MDS/MPN or MPN unclassified and decisive diagnostic assays such as cytogenetic analysis, FISH analysis or specific RT-PCR were not performed or only with delay. Moreover, newly available NGS technologies such as targeted RNA-sequencing, whole transcriptome or whole genome sequencing have revealed an increasing number of cytogenetically cryptic [51] or cytogenetically difficult to identify fusion genes, e.g., ETV6::ABL1 and several fusion genes with involvement of PDGFRB [52].
A rarely recognized feature of MLN-TK is monocytosis >1.0 × 10 9 /l which was identified in about one third of patients. It was clearly clustered in patients with hypereosinophilia >1.5 × 10 9 /l and consequently in patients with FIP1L1::PDGFRA or ETV6::ABL1 fusion genes. However, only approximately 20% of these patients also had relative monocytosis ≥10%. Even with taking into account the new cut-off values for diagnosis of chronic myelomonocytic leukemia with absolute monocytosis of ≥0.5 × 10 9 /l and relative monocytosis of ≥10%, these numbers did not substantially change. The data, therefore, clearly indicate that a MLN-TK may cause monocytosis but accompanying features include hypereosinophilia and relative monocytosis <10% in the vast majority of patients (Table 5).
Besides MLN-TK, the concurrent presence of significant eosinophilia and monocytosis is also a typical feature in patients with advanced systemic mastocytosis [53,54]. While being the disease-defining characteristic for chronic myelomonocytic leukemia, monocytosis is also identified in other myeloid neoplasms, potentially as marker of poor prognosis, e.g., in polycythemia vera, myelofibrosis and systemic mastocytosis [55,56]. These data therefore also underscore the current guidelines for diagnosis of chronic myelomonocytic leukemia, other myelodysplastic/myeloproliferative neoplasms and myeloproliferative neoplasms unclassified that the primary genetic work-up should not only exclude BCR::ABL1 positive chronic myeloid leukemia but also cases of MLN-TK. Due to the excellent prognosis of imatinib-treated patients with PDGFRA/PDGFRB [5,57,58] fusion genes, monocytosis had no obvious impact on progression and survival.
FIP1L1::PDGFRA and ETV6::ABL1 fusion genes share striking clinical and morphological similarities including male predominance, the relative frequency of eosinophilia and monocytosis, the median absolute number of eosinophils, and presentation or progression to BP including EMD. Of interest, progression to lymphoid BP in the BM seems to be a rare event for both fusion genes. Compared to FIP1L1::PDGFRA, the responses of ETV6::ABL1 positive patients to imatinib, nilotinib or dasatinib are less frequent and less durable [16]. In the current update of our own cohort, 6/11 patients were initially treated with imatinib but more durable remissions were only observed on primary or secondary treatment with nilotinib (n = 2), dasatinib (n = 3) or after allogeneic SCT (n = 1). Not included in our series, but important to note is that the MLN-TK subcategory also includes very rare fusion genes with involvement of other TK such as FLT3 [52,59].   In summary, the relative frequency of the defining characteristics of MLN-TK such as myeloid or lymphoid BP and/or eosinophilia occur at markedly variable frequencies according to the underlying fusion gene. Monocytosis is a potentially important marker which frequently occurs in association with significant eosinophilia. Careful attention must be paid to these subtle characteristics to avoid missing a diagnosis of a TKI-sensitive MLN-TK.

DATA AVAILABILITY
Freely available to any researcher