Chronic myelomonocytic leukemia (CMML) is a myeloid neoplasm characterized by sustained peripheral blood monocytosis (absolute monocyte count ≥1x10⁹/L, with monocytes accounting for ≥10% of the white blood cells), predominantly arising in the context of age-related clonal hematopoiesis, with overlapping features of myelodys-plastic syndromes (MDS) and myeloproliferative neoplasms (MPN).^1-3 The exact incidence and prevalence rates for CMML are hard to define, with Surveillance, Epidemiology, and End Results (SEER) registry data demonstrating an incidence of 0.4 cases per 100,000, with most studies showing a clear male preponderance.^4-7 The median age of presentation for CMML patients is between 70-75 years, with “young CMML” patients being operationally defined as those who present at <65 years of age.^4-8.

At the genome level CMML is relatively homogeneous, demonstrating approximately 10-12 somatic variants per kilobase of coding region, with most pathogenic variants involving TET2 (60%), ASXL1 (40%), SRSF2 (50%) and RAS pathway (30%) genes. However, clinically the disease is very heterogenous in presentation and outcomes, making diagnostic, prognostic and therapeutic decision-making challenging.^2,3,9-11 Broadly, CMML can be classified into dys-plastic CMML (dCMML), presenting with cytopenias and clinical signs and symptoms related to the same (fatigue, bruising and transfusion dependence) and proliferative CMML (pCMML), presenting with significant myeloproliferation, extramedullary hematopoiesis and associated constitutional symptoms (fever, weight loss, night sweats, anorexia, pruritus, bone pain and cachexia).¹⁰ From a classification perspective, for several years CMML was classified as a subtype of MDS, with the World Health Organization (WHO) rightfully and formally classifying CMML as an MDS/MPN overlap neoplasm, from 2002 onwards.¹ In 2015, the International Working Group for MDS/MPN overlap neoplasms proposed CMML-specific disease response criteria, providing support for the recognition of CMML as a specific disease entity and providing impetus for CMML-specific clinical trials.¹² These changes have clearly incentivized the development of disease-specific diagnostic, prognostic, and therapeutic strategies for patients with CMML. In this review, I discuss my approach to the diagnosis, prognosis, and management of patients with CMML.

CMML is a neoplasm associated with aging, often arising in the background of clonal hematopoiesis (bi-allelic TET2, or TET2/SRSF2 mutations), with the subsequent acquisition of mutations involving signaling (RAS pathway or JAK2V617F), epigenetic regulation (SETBP1, DNMT3A and EZH2), transcription factors (RUNX1) and pre mRNA splicing (SF3B1 and U2AF1), shaping clinical phenotypes (Figure 1).^2,3,13 There are pCMML subtypes in which oncogenic RAS pathway mutations (NRAS, CBL, KRAS and PTPN11) are clear initiating driver mutations, occurring early in the course of disease, associated with poor outcomes (decreased survival and higher rates of transformation to acute myeloid leukemia [AML]) (Figure 1).^10,14 Unlike in MDS, MPN or AML, TP53 mutations are extremely infrequent in CMML (<1%), and are only really encountered at the time of CMML to AML transformation, or in the context of therapy-related CMML.^11,15,16

The 2016 iteration of the WHO classification of myeloid neoplasms has outlined diagnostic criteria for CMML which include the presence of sustained peripheral blood monocytosis, absence of reactive causes, the presence of <20% blasts and promonocytes (blast equivalents) in the peripheral blood and bone marrow, and exclusion of molecularly defined myeloid neoplasms that can present with monocytosis (BCR-ABL1, PDGFRA, PDGFRB, FGFR1 and PCM1-JAK2 rearrangements), with or without dysplasia (Table 1).¹ In the absence of dysplasia, a diagnosis of CMML can be made if the monocytosis has persisted for ≥3 months, reactive causes have been excluded, or somatic cytogenetic or molecular markers frequent in CMML (e.g., ASXL1, TET2, SRSF2 and SETBP1) can be documented. While this approach is very reasonable, there are limitations and important nuances associated with these criteria.

While absolute monocytosis is uncommon in chronic myeloid leukemia, it can occur in BCR-ABL1 p190 isoformdriven disease and, regardless of the presence or absence of a “myelocyte bulge”, assessment for BCR-ABL1 fusions by fluorescence in situ hybridization, cytogenetics, and/or molecular techniques should be pursued.¹⁷ Chromosomal translocations/rearrangements involving PDGFRA and PDGFRB can give rise to myeloid neoplasms, often characterized by prominent eosinophilia and responsiveness to imatinib.^18-20 Among these, PDGFRB-rearranged myeloid neoplasms can be associated with absolute monocytosis (<1% of all cases morphologically diagnosed as CMML), usually with concomitant eosinophilia, and given their unique responsiveness to imatinib are best classified as molecularly defined neoplasms and not as CMML.^18-20 More than 20 fusion partner genes have been described with PDGFRB, with t(5;12)(q31-q32;p13), giving rise to the ETV6(TEL)-PDGFRB fusion, being the most common.²¹ The FIP1L1-PDGFRA fusion arising due to the CHIC2 deletion is the most common PDGFRA aberration and is uncommonly associated with monocytosis.¹⁹ While most PDGFRB re-arrangements can be identified by conventional karyotyping, the FIP1L1-PDGFRA fusion is karyotypically occult and can only be detected by fluorescence in situ hybridization or molecular analyses. Similarly, FGFR1 and PCM1-JAK2 rearrangements are very uncommon causes of monocytosis and are more commonly associated with eosinophilia. Monocytosis can occur in the context of other myeloid neoplasms such as MDS and MPN and is associated with poor outcomes in MPN.^22,23 While monocytosis in MDS can be a reflection of an ongoing evolutionary trajectory to CMML (oligo-monocytic CMML), in MPN, the utilization of monocyte repartitioning flow cytometry (discussed below) and driver mutation status can help differentiate CMML from MPN with monocytosis.^23,24 Among classical MPN-driver mutations, while JAK2V617F can occur in 10% of CMML patients, mutations involving MPL and CALR are extremely infrequent and their detection should raise questions with regards to a bona fide CMML diagnosis.^3,25 Occasionally NPM1 and FLT3 driver mutations are identified in CMML patients with excess blasts (5-19%).^26,27 For all practical purposes I consider these cases as acute myelomonocytic leukemia in evolution and treat them as such.

Reactive monocytosis is very common in practice and while viral infections and recovering bone marrow (from injury, drugs or chemotherapy) are frequent causes, sustained reactive monocytosis is more common in chronic infections such as subacute bacterial endocarditis, tuberculosis, brucellosis, leishmaniasis and leprosy and in autoimmune/inflammatory disorders such as systemic lupus erythematosus, sarcoidosis and mixed connective tissue disorder.²⁸ Reactive monocytosis can also be seen in the context of metastatic visceral neoplasms, either due to enhanced mobilization of monocytes from the bone marrow, or due to increased monopoiesis mediated by CCL2 (C-C motif chemokine ligand 2).²⁹

Conventional flow cytometry has a limited diagnostic role in CMML, given that immature/neoplastic monocytes do not express unique surface markers and that promonocytes/monoblasts are frequently CD34-negative (blast marker).³⁰ Flow abnormalities that can be detected in CMML include abnormal myeloid maturation patterns involving CD11b, CD13 and CD16, along with aberrant expression of CD56 on monocytes.³⁰ Monocyte repartitioning by flow cytometry has gained popularity, especially given its ability to differentiate CMML from other reactive and clonal causes of monocytosis.^31-33 Based on the expression of CD14 and CD16, monocytes can be divided into three categories: CD14⁺/CD16^- classical (M01), CD14^low/CD16⁺ intermediate (M02), and CD14^-/CD16⁺ non-classical monocytes (M03) (Figure 2).^31,32 These subsets differ in their chemokine receptor expression, phagocytic activity, epigenetic profiles and have unique metabolic pathway dependencies.^32,34 In CMML, a pivotal study demonstrated an increase in the M01 subset, with an established cutoff >94% being associated with sensitivity and specificity values of 90.6% and 95.1%, respectively.³² These findings have been validated independently and this method importantly is effective in identifying MDS patients whose disease eventually evolves into CMML and in distinguishing CMML from MPN with monocytosis.^31,33,35,36 False negative findings secondary to autoimmunity/inflammation (expansion of the M02 fraction) and false positive findings in myeloid neoplasms such as MDS, atypical chronic myeloid leukemia and classical chronic myeloid leukemia have been documented.^33,37,38 I do use this flow cytometry assay for screening patients who present with sustained monocytosis, especially when there is suspicion of an underlying clonal process and to differentiate CMML-associated monocytosis from other myeloid neoplasms with monocytosis.³⁹

Next-generation sequencing assays are an important part of the diagnosis and prognostication of CMML. Virtually all CMML patients will have detectable somatic mutations involving genes regulating the epigenome, splicing, signaling and transcription.^4,5,11,40 Clonal compositions help to define pCMML and dCMML subtypes and contribute towards AML transformation.¹⁰ Single-cell sequencing data have shown that TET2 mutations are usually the founder mutations occurring at the hematopoietic stem cell level.⁴¹ These mutations impact multipotent progenitor and common myeloid progenitor cells, skewing differentiation towards granulomonocytic progenitors and mature monocytes, respectively (Figure 1).⁴¹ Second-order mutations tend to accumulate in multipotent progenitor and common myeloid progenitor cells, often involving additional sites/alleles on TET2, spliceosome component genes (SRSF2, rarely SF3B1 and U2AF1) and additional epigenetic regulators (ASXL1, rarely EZH2).^11,41-43 Signaling mutations, such as NRAS, CBL, PTPN11, KRAS, NF1 and JAK2V671F, can sometimes be early/founder mutations, with inherent hypersensitivity of hematopoietic stem/progenitor cells to granulocyte-macrophage colony-stimulating factor (like the pCMML pediatric counterpart, juvenile myelomonocytic leukemia), but can also be later/subclonal events, giving rise to pCMML.^10,41,44 Similarly, acquisition of additional epigenetic (DNMT3A, rarely IDH1/2) and splicing mutations (SF3B1) and mutations involving transcription factors (RUNX1) often gives rise to a dCMML phenotype.^40,45 Somatic copy number alterations, especially copy neutral loss of heterozygosity, is common in CMML and frequently involves TET2 (4q24) and CBL (11q23), playing a role in clonal evolution.¹⁰ Genetic alterations involving protein coding regions, including copy number gains and losses in driver mutations were only able to explain 44% of CMML cases that transformed to AML, indicating that mechanisms of AML transformation remain to be elucidated.¹⁰ It is important to note that germline variants have been implicated in the pathogenesis of CMML, with cases having been documented in the context of germline mutations involving RUNX1, ANKRD26, ETV6, CHEK2, CDK2NA and GATA2.^46,47 Recently, a 700 kb germline duplication localized to 14.q32.2, resulting in overexpression of ATG2B and GSKIP genes, was described and has been associated with familial MPN and CMML.⁴⁸ Based on family histories, age of onset and the heterozygous nature of variant allele fractions, I routinely assess germline DNA from extracted hair follicles/skin fibroblasts to assess for germline predisposition syndromes.⁴⁹

In CMML, bone marrow biopsies are often hypercellular with granulocytic hyperplasia and mild to modest dysplasia (Figure 3D, E). Bone marrow monocytosis can be present, but is often difficult to appreciate and immunohistochemical studies that aid in the identification of monocytes and their precursors are recommended (Figure 3C).⁵⁰ Almost 80% of patients will demonstrate micro-megakaryocytes with abnormal nuclear contours and lobations, and 20-30% of patients can have an increase in reticulin fibrosis.⁵⁰ Approximately 30% of patients demonstrate nodules composed of mature plasmacytoid dendritic cells that are clonal (CD123⁺,lineage-negative, CD45⁺, CD11c^–, CD33^–, HLA-DR⁺, BDCA-2⁺ and BDCA-4⁺), often have RAS pathway mutations and predict for an inferior AML-free survival (Figure 3F).⁵¹ The identification of promonocytes requires expertise and these cells should be summated with blasts when estimating the blast count (Figure 3A, B).⁵² Promonocytes are described as monocytic precursors that have a delicately convoluted, folded or grooved nucleus with finely dispersed chromatin, a small indistinct or absent nucleolus, and finely granulated cytoplasm.^52,53 On immunophenotyping the abnormal bone marrow cells often express myelomonocytic antigens such as CD13 and CD33, with variable expression of CD14, CD68 and CD64. Markers of aberrant expression include CD2, CD15, and CD56 or decreased expression of CD13, CD14, HLA-DR, CD64 or CD36. The presence of myeloblasts can often be detected by expression of CD34. The most reliable markers on immunohistochemistry include CD68R and CD163. On cytochemical analysis, monocytes are often positive for non-specific esterases and lysozyme, while the granulocytic precursors are often positive for lysozyme and chloroacetate esterase (Figure 3C). This technique can help to differentiate CMML from other overlap neoplasms in which bone marrow monocytosis is uncommon. Conventional karyotyping on the bone marrow is important for cytogenetic risk stratification, with approximately 70-80% of patients demonstrating a normal karyotype. Common abnormalities include +8 and –Y, with isolated del5q, complex and monosomal karyotypes being very uncommon (complex and monosomal karyotypes can be seen in patients with therapy-related CMML).^16,54,55 The CMML-specific prognostic scoring system (CPSS) and the Mayo-French cytogenetic risk stratification system are two commonly used karyotype-based prognostic models for patients with CMML (Table 2).^54,55

The diagnosis and management of CMML variants and molecularly defined entities presenting with monocytosis, mimicking CMML, require special attention. These variants can broadly be divided into three categories: (i) oligomonocytic CMML; (ii) CMML associated with a concomitant myeloid neoplasm; and (iii) molecularly defined entities with monocytosis.

Oligomonocytic CMML. This category encompasses patients who present with sustained relative monocytosis (≥10% of white blood cells) and absolute monocytosis not meeting current diagnostic criteria for CMML (absolute monocyte count 0.5-<1.0x10⁹/L).²⁴ Based on the 2016 WHO classification, these patients would be classified as having either MDS or MDS/MPN-Unclassifiable.¹ Except for an absolute monocyte count of ≥1x10⁹/L, if these patients meet other CMML diagnostic criteria, along with a M01 fraction >94% on monocyte repartitioning flow cytometry, and a molecular signature consistent with CMML (TET2, SRSF2, and ASXL1), I consider them as having oligomonocytic CMML and follow and manage them as such. Over time, several of these patients will have clonal evolution to either CMML or secondary AML.

CMML associated with a concomitant myeloid neoplasm. This category includes several variants, with the two most important being systemic mastocytosis (SM) with CMML (SM-CMML) and JAK2V617F-mutant CMML²⁴ CMML is the most frequent hematologic neoplasm associated with SM. I diagnose SM-CMML in patients who meet WHO criteria for both entities. In these patients, the SM component can present as either indolent or aggressive SM, with mast cell leukemia being extremely infrequent.⁵⁶ In a Mayo Clinic study of 50 patients with SM-CMML, survival outcomes were similar to those of patients with CMML (24 months for SM-CMML vs. 18 months for CMML; P=0.08), There was a higher frequency of KIT and CBL mutations in SM-CMML and CMML-based prognostic models were not effective in risk stratification in this condition due to the confounding impact of SM.⁵⁶ In CMML patients who have a detectable KITD816V mutation, a known driver oncogene in SM (>90% of cases), I universally assess for concomitant SM clinically and by using laboratory techniques such as serum tryptase levels (>20 ng/mL), bone marrow morphology, flow cytometry (aberrant expression of CD2 and/or CD25 on mast cells) and immunohistochemistry (CD117 and tryptase). For SM-CMML patients there are several exciting KIT-directed targeted therapies for the SM component (e.g., midostaurin and avapritinib) and given that in SMCMML, neoplastic monocytes are also often KIT mutated,⁵⁷ these drugs can sometimes effectively decrease monocyte counts and infiltrative disease burden.⁵⁸ JAK2V617F-mutant CMML usually gives rise to a pCMML subtype with higher hemoglobin levels and absolute monocyte counts, with monocyte repartitioning by flow being a useful tool to distinguish this entity from MPN with monocytosis.⁵⁹ On rare occasions, it becomes very difficult to distinguish the concomitant presence of a JAK2V617F MPN with bona fide CMML and in these instances, I manage each symptomatic component individually.

Molecularly defined entities with monocytosis. This category includes myeloid and lymphoid neoplasms presenting with monocytosis in the context of rearrangements involving PDGFRA, PDGFRB, FGFR1 and PCM1-JAK2. These entities are best designated by their molecular drivers and are currently not referred to as CMML. The recognition of these entities is very important given that they can present with a lymphoproliferative component and eosinophilia and can be targeted with tyrosine kinase inhibitors.^19,60

While there are numerous prognostic models for patients with CMML, models such as the Bournemouth, Lille, International Prognostic Scoring System (IPSS) and revised-IPSS scores are primarily designed for patients with MDS and excluded patients with pCMML.^61,62 The MD Anderson prognostic system (MDAPS) is CMML-specific and identified a hemoglobin level <12 g/dL, presence of circulating immature myeloid cells (myelocytes, promyelocytes and metamyelocytes), absolute lymphocyte count >2.5x10⁹/L and ≥10% bone marrow blasts as independent predictors for inferior survival.⁶³ The Global MDAPS was then developed for patients with MDS, secondary MDS and CMML, with prognostic factors including; older age, poor performance status, thrombocytopenia, anemia, increased bone marrow blasts, leukocytosis (>20x10⁹/L), chromosome 7 or complex cytogenetic abnormalities and a prior history of red blood cell transfusions.⁶⁴ This model identified four prognostic groups with median survivals of 54 (low), 25 (intermediate-1), 14 (intermediate-2) and 6 months (high), respectively.⁶⁴ The CPSS model was developed in Europe and identified pCMML versus dCMML subtypes, WHO CMML-subtypes, red blood cell transfusion dependency and the CPSS cytogenetic risk stratification system as being prognostic for survival.^54,65 The Mayo prognostic model identified hemoglobin <10 g/dL, platelet count <100×10⁹/L, absolute monocyte count >10×10⁹/L and circulating immature myeloid cells as being independently prognostic.⁶

The discovery of somatic mutations in CMML resulted in the development of contemporary molecular prognostic models. The Group Francophone des Myelodysplasies (GFM) demonstrated an adverse prognostic effect for truncating ASXL1 mutations in 312 patients with CMML; additional risk factors included age >65 years, white blood cell count >15×10⁹/L, platelet count <100×10⁹/L and hemoglobin <10 g/dL in females and <11 g/dL in males.⁵ The GFM model assigns three adverse points for white blood cell count >15×10⁹/L and two adverse points for each one of the remaining risk factors, resulting in a three-tiered risk stratification; low (0–4 points), intermediate (5–7) and high (8–12), with respective median survivals of 56, 27.4 and 9.2 months.⁵ To further clarify the prognostic relevance of ASXL1 mutations, the Mayo Molecular Model (MMM) was developed as a collaborative effort between the GFM and Mayo Clinic (n=466).⁶⁶ Adverse prognostic factors included truncating ASXL1 mutations, absolute monocyte count >10×10⁹/L, hemoglobin <10 g/dL, platelets <100×10⁹/L and circulating immature myeloid cells. Based on these variables a regression coefficient-based prognostic model was developed with the following risk categories; high (≥3 risk factors), intermediate-2 (2 risk factors), intermediate-1 (1 risk factor), and low (no risk factors) risk, with median survivals of 16, 31, 59 and 97 months, respectively.⁶⁷ The CPSS model was also updated to include gene mutations involving ASXL1, RUNX1, NRAS and SETBP1 (CPSS-Mol).⁴ Gene mutations along with karyotypic abnormalities are used to calculate the CPSS genetic score. One point each is assigned for an intermediate-1 genetic score, white cell count ≥13x10⁹/L, bone marrow blasts ≥5% and red blood cell transfusion dependency, two points for intermediate-2 genetic score and three points for a high risk genetic score.⁴ The CPSS-Mol stratifies patients into four categories, low (0 risk factors), intermediate-1 (1 risk factor), intermediate-2 (2-3 risk factors) and high (≥4 risk factors) risk, with median overall survival not reached, 64, 37 and 18 months; with 4-year leukemic transformation rates of 0%,3%, 21% and 48%, respectively.⁴ Table 2 highlights relevant CMML-specific prognostic models along with their component variables. Recently a CMML transplant model was developed (CMML transplant score) which assigned four points for the presence of ASXL1 and/or NRAS mutations, four points for bone marrow blasts >2% and one point for each hematopoietic stem cell transplant (HSCT) comorbidity index, effectively stratifying for both overall survival and non-relapse mortality.⁶⁸

In practice, any of the three molecularly integrated CMML prognostic models can be used for risk stratification. While these models have not been formally compared against each other, they share several overlapping prognostic features, especially anemia, elevated white blood cell counts and truncating ASXL1 mutations.^4,5,11 I use both the MMM and the CPSS-Mol model for risk stratification at my institution.

The first step in the management of CMML patients is establishing an accurate diagnosis, followed by personalized risk stratification. Using any one of the molecularly integrated prognostic models, CMML patients can be stratified into lower risk and higher risk groups (Figure 4). Management strategies for these two risk groups are described below.

CMML is a unique MDS/MPN overlap neoplasm with relative genetic homogeneity, but with marked clinical heterogeneity. The disease is seen in the elderly and frequently develops on the background of clonal hematopoiesis, with recurrent somatic mutations involving TET2, ASXL1, SRSF2 and the RAS pathway defining dysplastic and proliferative subtypes of CMML. For several years CMML was considered as a subtype of MDS, but from 2002 onwards, CMML has been rightfully recognized as a unique neoplasm with the development of CMML-specific prognostic models, response criteria, preclinical models and, most importantly, clinical trials; heralding a new future for this disease and affected patients. Several challenges remain, including the lack of uniform consensus on personalized prognostication and more importantly, the identification of disease-modifying targets and therapies that might ameliorate disease progression, improve quality of life, and potentially offer a cure.