The Molecular Features of Chronic Eosinophilic Leukemia, Not Otherwise Specified

Chronic eosinophilic leukemia, not otherwise specified (CEL-NOS), is an extremely rare type of myeloid neoplasm, belonging to a subclass of myeloproliferative neoplasms (MPNs). Due to the lack of specific molecular markers, it remains a diagnosis by exclusion. Although previous next-generation sequencing (NGS) studies on hypereosinophilic syndrome (HES) have included subsets of CEL-NOS patients, no studies to date have specifically focused on consecutively diagnosed CEL-NOS cohorts.

Here, we summarize the clinical and laboratory data of 15 consecutively diagnosed patients with CEL-NOS, who referred to our hospital between July 2019 and October 2024. Diagnosis of CEL-NOS were reviewed and reclassified according to the 2022 World Health Organization (WHO) criteria [1]. Peripheral eosinophils ≥ 10% of leukocytes and an absolute eosinophil count (AEoC) ≥ 1.5 × 10⁹/L were observed in all CEL-NOS patients. Protein-coding region analysis was performed on bone marrow (BM) samples from all patients using a targeted panel of 137 hematolymphoid tumor-associated genes (see Data S1: Supplementary methods). A cohort of 65 patients with Idiopathic HES (IHES) diagnosed during the same period was collected as a control group. Statistical analysis is based on the clinical and laboratory parameters collected at the time of referral (see Data S1: Supplementary methods).

Compared to patients with IHES, CEL-NOS patients have a higher median age at presentation. The primary clinical manifestations at presentation are splenomegaly and constitutional symptoms, while skin, digestive, and respiratory symptoms are relatively uncommon (Table 1). Patients with CEL-NOS have higher WBC and lower hemoglobin and platelet concentrations. Although the absolute eosinophil and basophil counts in the complete blood count (CBC) are higher in the CEL-NOS group, there is no significant difference in their relative proportions between the two groups. CEL-NOS patients have higher serum levels of vitamin B12 and lactate dehydrogenase (LDH), as well as lower immunoglobulin E (IgE) levels. There is no significant difference in serum interleukin-5 (IL-5) levels between the two groups. In each group, clonal T-cell receptor (TCR) rearrangements were detected in 2 patients, but no abnormal lymphocyte immunophenotypes were found by multi-parameter flow cytometry (MPFC). Patients with CEL-NOS have increased BM cellularity and myeloid-to-erythroid (M:E) ratios compared to those with IHES. Among the 15 CEL-NOS patients, only one (6.7%) had a BM blast percentage of ≥ 5% at the time of presentation; however, fibrosis of grade ≥ 2 was observed in 5 patients (33.3%).

Cytogenetic abnormalities were detected in 8 (53.3%) CEL-NOS patients, with trisomy 8 and del(20q) identified in 3 (20%) and 2 (13.3%) cases, respectively. Other abnormalities were detected in single cases (see Tables 1 and S2). In comparison, among 65 IHES patients, only one case demonstrated an abnormal karyotype of 45,XY,-Y[20]. This patient had no other abnormalities in CBC or BM morphology and was sensitive to glucocorticoid treatment. A panel of 137 genes revealed that all CEL-NOS patients had at least one pathogenic mutation (see Figure 1 and Table S2). The most common recurrent genetic abnormalities in CEL-NOS were STAT5B, followed by ASXL1, TET2, SETBP1, NRAS, SRSF2, and KIT. Among the three patients with the KIT D816V mutation, no definitive evidence of abnormal mast cell proliferation was found through immunohistochemical staining and MPFC. The identification of cytogenetic and molecular abnormalities plays a pivotal role in differentiating CEL-NOS from IHES. However, the potential presence of clonal hematopoiesis of indeterminate potential (CHIP) in IHES patients must be carefully considered. NGS analysis identified clonal hematopoiesis in 13 (20%) of IHES patients. Notably, these cases predominantly exhibited aging-related genetic abnormalities, typically with variant allele frequencies (VAF) below 10%. The recurrent mutations identified were limited to DNMT3A (n = 7) and TET2 (n = 2), with a median VAF of 1.94% (interquartile range [IQR], 1.3%–4.7%) and 5.15%, respectively. Single cases were identified for each of the following seven genes: ASXL1 (VAF: 4%), ARID1A (4.1%), BCOR (13.2%), JAK2 (2%), IDH2 (2.1%), KMT2D (2.6%), and ZRSF2 (2.5%). Except for TET2, no overlapping recurrent genetic alterations were observed between the two groups. Notably, CEL-NOS patients with TET2 mutations exhibited a higher median VAF value of 50.1% (IQR, 42.4%–76.09%); the lower median VAFs of CHIP-related genes in the IHES cohort may be associated with their younger median age compared to the CEL-NOS group.

Cross et al. found that STAT5B N642H is associated with myeloid neoplasms with eosinophilia, and this finding has been confirmed by several recent studies [2-4]. In the CEREO study, Groh et al. demonstrated that 54% of 64 patients with HES/HE exhibiting myeloid neoplasm features harbored at least one JAK–STAT signaling pathway abnormality [4]. Consistent with these findings, our data revealed JAK–STAT pathway gene abnormalities in 10 of 15 (66.7%) CEL-NOS patients. However, the STAT5A mutation recently identified by Groh et al. was not included in our gene panel analysis. Among the 8 cases with STAT5B mutations, 7 (87.5%) represented dominant clones with a median VAF of 42.4% (IQR, 35.7%–68.7%). The STAT5B mutations included N642H (n = 6), T628S (n = 1), and I704L (n = 1). Two patients harbored distinct JAK1 mutations: R629_D630del and V658F. In addition to the STAT5B I704L mutation (VAF: 46.8%), all other mutations have been documented in CEL-NOS/HES cases. Verification using oral mucosal cells identified this mutation as somatic. To elucidate the hematological characteristics of other myeloid neoplasms with STAT5B mutations, we analyzed NGS data from over 2300 myeloid neoplasm patients during the same period. STAT5B mutations were identified in 59 patients, including 7 cases of AMLs, 29 of myelodysplastic syndromes (MDSs), 16 of MPNs, 6 of myelodysplastic/myeloproliferative neoplasms (MDS/MPNs), and 1 of myeloid neoplasm with PCM1::JAK2 fusion. Notably, only three of these patients exhibited an AEoC of ≥ 1.5 × 10⁹/L, but these three patients were accompanied by CBFβ::MYH11, PCM1::JAK2, and JAK2 V617F rearrangements, respectively. These findings suggest that STAT5B mutations alone are insufficient to drive eosinophilia. Rather, the disease phenotype may result from the combined effects of additional epigenetic regulators and RNA splicing factors.

To the best of our knowledge, this study represents the first report identifying STAT5B as the most frequently mutated gene in CEL-NOS. Consistent with previous findings, STAT5B frequently functions as the dominant clone, with the N642H mutation being the predominant variant [3, 4]. However, it should be noted that STAT5B mutations alone are insufficient to induce the eosinophilia phenotype, and the precise pathogenic mechanisms underlying this condition warrant further investigation. In the 2022 WHO classification, the percentage of blasts was removed as clonal evidence for CEL-NOS [1]. Our research found that at least one pathogenic mutation could be detected in all 15 patients with CEL-NOS, while only patient E3 had a BM blast of ≥ 5% at diagnosis. None of the patients received a CEL-NOS diagnosis based exclusively on blast count as the sole evidence of clonality. Therefore, the removal of blast percentage does not impact the diagnosis of CEL-NOS.

After a median follow-up of 20 months (IQR, 13–35) from illness onset (10 months from the diagnosis), eight deaths (53.5%) were recorded. The median survival from diagnosis to death for CEL-NOS patients was 26 months (IQR, 13–35). Among the eight deceased patients, mortality was attributed to acute myeloid leukemia (AML) transformation in four cases, while single cases resulted from cerebral hemorrhage, transplantation-related complications, COVID-19 infection, and an unidentified cause, respectively. Of the four patients who progressed to AML, only patient E14 received treatment at our institution. This patient developed AML 6 months following the initial diagnosis and was treated with multiple therapeutic regimens, including decitabine, DA 3 + 7 protocol, and azacitidine in combination with venetoclax. Despite these interventions, the patient failed to achieve remission and succumbed to infection 4 months post-transformation. Among the seven surviving patients, three achieved partial hematological remission following combination therapy with interferon and glucocorticoids, maintaining therapeutic responses for 4–9 months. Two patients received hydroxyurea as primary cytoreductive therapy. One patient presenting with elevated bone marrow blast counts achieved complete hematological remission after 6 cycles of azacitidine therapy. Notably, patient E10, harboring both the JAK1 R629_D630del mutation and trisomy 8 chromosomal abnormality, was initially treated with ruxolitinib (15 mg orally twice daily). However, after 1 month of therapy, no reduction in AEoC was observed. Following treatment modification to pegylated interferon-α2b, complete hematological remission was achieved within 2 months. Our research has indicated that patients suffering from CEL-NOS generally exhibit poor clinical outcomes. Following diagnosis, immediate evaluation for suitable HLA-matched donors is recommended, accompanied by rigorous disease monitoring. Allogeneic transplantation should be promptly considered upon disease progression as the potentially curative therapeutic intervention.

Several limitations should be acknowledged in our study. First, potential selection bias may exist due to the tertiary referral nature of our medical center, which typically manages more severe disease presentations. This is evidenced by the observed transformation rate of CEL-NOS to acute leukemia in our cohort (27%), which aligns with the elevated rates (24%–50%) reported in previous studies from similar tertiary centers [5]. However, this contrasts markedly with the < 1% transformation rate reported in a population-based study of 373 CEL-NOS patients [6]. Second, comprehensive treatment data were limited as most patients subsequently received care at local hospitals or through home-based treatment following the initial diagnosis.

In conclusion, our study demonstrates that CEL-NOS is characterized by complex molecular profiles, with the STAT5B mutation as the most prevalent genetic alteration. Given that our study was conducted in a single center with a limited sample size, further research is needed to validate our findings.

Z.J.X. designed the research, was the principal investigator, and took primary responsibility for the paper; S.Q.Q., N.N.L., T.J.Q., Z.F.X., B.L., L.J.P., Q.Y.G., M.J., and Z.J.X. recruited the patients; Q.S., H.J.W., and Y.J.J. contributed to pathological diagnosis; S.Q.Q. and N.N.L. acquired the clinical data; S.Q.Q. and Z.J.X. analyzed data and wrote the manuscript.

This study was approved by the Ethics Committee of Blood Disease Hospital, Chinese Academy of Medical Sciences, compliant with the principles of the Declaration of Helsinki. Patients gave written informed consent.

The authors declare no conflicts of interest.