Advances in chronic myelomonocytic leukemia and future prospects: Lessons learned from precision genomics

Abstract

In the latest World Health Organization classification of myeloid neoplasms, chronic myelomonocytic leukemia (CMML) exists as a separate entity under the category of myelodysplastic/myeloproliferative (MDS/MPN) overlap syndromes. Outcomes remain uniformly poor with a median overall survival of ~2 years and an inherent risk of transformation into acute myeloid leukemia (15–20% over 5 years). Due to unique biologic characteristics such as overlapping features of myelodysplasia and myeloproliferation, and clinical diversity despite relative genomic homogeneity, CMML represents a unique model to study chronic myeloid tumor biology. Recent advances have focused on understanding the role of putative genomic abnormalities, in particular, clonal evolution of pathogenic alterations in genes regulating the epigenome (TET2), chromatin architecture (ASXL1), spliceosome complex (SRSF2, SF3B1) and cell signaling (NRAS, KRAS, CBL, JAK2). Disease prognostication has evolved from purely clinical prognostic models to those incorporating pathogenic gene variations. Therapeutic options in this disease remain dismal with only two agents approved by the United States Food and Drug Administration, namely 5-azacitidine and decitabine. Allogeneic hematopoietic stem cell transplantation remains the sole curative option in this disease; however is associated with substantial treatment-related morbidity and mortality. Future areas of research include opportunities to further improve disease prognostication by employing novel technologies such as machine learning, incorporation of methylation and cytokine signatures, in addition to gene mutations; insights into clonal origins of this disease, and novel therapeutic strategies.

Introduction:

As per the 2016 World Health Organization (WHO) criteria¹, chronic myelomonocytic leukemia (CMML) is classified as a myelodysplastic/myeloproliferative (MDS/MPN) overlap syndrome, with overlapping features of myelodysplasia and myeloproliferation². Being primarily a disease of aging with a median age of diagnosis ~70–75 years^3–5, there are approximately 4 CMML cases per 100, 000 persons annually in the United States of America^4,6,7, with a male preponderance (M:F ratio: 1.5 to 3:1) and an inherent tendency (~15–20% over 3–5 years) towards leukemic transformation⁸.

Diagnosis is established by the presence of persistent (>3 months) peripheral blood (PB) monocytosis [(≥1 × 10⁹/L and monocytes comprising of ≥10% of the white blood cell (WBC) differential count] and absence of other disease-defining genetic abnormalities such as PDGFRA, PDGFRB, FGFR1 rearrangements, PCM-1 JAK2 and BCR-ABL1 translocations, among others^2,4,6. Cytogenetic abnormalities are seen in ~one-third cases (most commonly trisomy 8, followed by chromosome 7 deletion or rearrangements⁹), while well-established pathogenic gene variants are seen in >90% cases^10–12. In spite of the high frequency of gene mutations, there are no mutations specific to this disease¹³. However, compared to cancers with high mutational burden such as melanoma with >1000 mutations per megabase of coding DNA (Mb), CMML is genetically more homogenous, with between 10–15 somatic mutations in coding sequences per exome^14,15. Majority of these mutations are associated with epigenetic dysregulation, primarily deoxyribonucleic acid (DNA) methylation and chromatin abnormalities, secondary to altercations in the ten-eleven translocation-2 (TET2) and additional sex coombs-like 1 (ASXL1)^12,16–19 genes, and splicing defects, predominantly involving the serine/arginine-rich splicing factor 2 (SRSF2)¹⁰. Along with mutations, there are other intriguing aspects to CMML biology such as hypersensitivity to granulocyte monocyte colony stimulating factor (GM-CSF)²⁰ and altered cytokine profiles²¹, which make this disease an attractive model to study not only clonal evolution in chromic myeloid neoplasia, but also therapeutic manipulation of the involved epigenomic, genomic and signaling pathway abnormalities. In this manuscript, we review the current prognostic models and their limitations, therapeutic options and provide future perspectives in towards understanding disease biology, prognosis and management in CMML.

Disease biology and genomic abnormalities:

I. Chromosomal abnormalities:

As iterated before, cytogenetic abnormalities are only present in ~one-third CMML patients, while pathogenic gene variants are present in a significant majority (>90%)^9,10,22. In a Spanish cohort of 414 CMML patients, the most frequent karyotypic abnormalities were found to be trisomy 8 (+8, 7.2%), isolated loss of chromosome Y, (-Y, 4.3%), abnormalities of chromosome 7, and complex (≥ 3 abnormalities) karyotype. Based on outcome analysis, three cytogenetic risk categories were identified with low risk (normal karyotype or loss of Y chromosome as a sole abnormality), high risk (trisomy 8, chromosome 7 abnormalities or complex karyotype) and intermediate risk (all other abnormalities)⁹. In a Mayo Clinic-French consortium cohort of 268 CMML patients, the most common chromosomal abnormality was +8 (23%), followed by -Y (20%), −7/7q- (14%), 20q- (8%), +21 (8%), and der (3q) (8%)²³. Risk stratification in this model included differences from the Spanish model, such as incorporation of trisomy 8 in the intermediate and sole der (3q) in the low risk category²³. This risk model stratified patients into three categories, high, intermediate and low with median overall survivals of 3, 20, and 41 months respectively²³.

II. Genetic and epigenetic alterations:

Age-related pathogenic gene signatures are frequent in CMML, with TET2 (60%) being the most frequent gene altered, followed by SRSF2 (50%) and ASXL1 (40%)⁴. Table 1 provides a list of known somatic genes altered in CMML, along with their chromosomal location, function, frequency and prognostic impact.

Table 1:

Table showing list of somatic gene alterations in chronic myelomonocytic leukemia (in order of decreasing frequency).

Gene name	Chromosomal location	Function	Frequency	Independent prognostic impact
TET2	4q24	DNA methylation	60%	No
SRSF2	17q25.1	Splicing	50%	No
ASXL1	20q11.21	Histone modification	40%	Yes
NRAS	1p13.2	Cell signaling	15%	Yes
CBL	11q23.3	Cell signaling	15%	No
RUNX1	21q22.12	Transcription and nucleosome assembly	15%	Yes
SETBP1	18q12.3	Transcription and nucleosome assembly	15%	Yes
KRAS	12p12.1	Cell signaling	10%	No
IDH2	15q26.1	Histone modification and DNA methylation	5–10%	No
JAK2V617F	9p24.1	Cell signaling	5–10%	No
SF3B1	2q33.1	Splicing	5–10%	No
U2AF1	21q22.3	Splicing	5–10%	No
EZH2	7q36.1	Histone modification	5%	No
DNMT3A	2p23.3	DNA methylation	5%	Yes
PTPN11	12q24.13	Cell signaling	5%	No
ZRSR2	Xp22.2	Splicing	5%	No
FLT3	13q12.2	Cell signaling	<5%	No
IDH1	2q34	Histone modification and DNA methylation	1%	No

The traditional model of clonal evolution in CMML assumes a stepwise accumulation of gene mutations with clonal expansion and selection (figure 1). Seminal work studying the clonal architecture through serial replating assays and genotyping of CMML patient-derived colony assays has suggested the early clonal dominance of mutations in TET2 and ASXL1, compared to transcription factor (RUNX1) and signaling pathway (RAS) mutations; however the order of acquisition was not found to be fixed. This contrasts to findings in juvenile myelomonocytic leukemia, the pediatric counterpart of adult CMML, where germline or somatic RAS pathway mutations are thought to drive oncogenesis²⁴. Preliminary work (unpublished) from our laboratory challenges the existing model of clonal hierarchy in CMML and suggests that RAS alterations are in fact, early clonal events, at least in the myeloproliferative subtype of this disease, while epigenetic alterations are secondary cellular events which consequently skew hematopoiesis towards a myeloid compartment bias and monocytosis.

Figure 1.

Figure showing a simplistic model to understand clonal evolution in CMML. TET2 mutation is thought to be an early event in a hematopoietic precursor (likely HSC), thus establishing early clonal dominance. This is followed by acquiring of a mutation in a gene regulating spliceosome complex, most commonly SRSF2. This skews hematopoiesis towards a myeloid compartment bias, thus resulting in monocytosis. Late clonal dominance is established either with a mutation in chromatin-modifying (ASXL1) or RAS (NRAS/KRAS/CBL/JAK2) pathway gene, thus resulting in dysplastic subtype in the former (MDS-CMML), and proliferative subtype (MP-CMML) in the latter. Abbreviations: HSC = hematopoietic stem cell; CMP = common myeloid progenitor; GM = granulocyte-monocyte; GMP = granulocyte-monocyte progenitor; MDS = myelodysplastic; MP = myeloproliferative; CMML = chronic myelomonocytic leukemia.

The sentinel genomic event responsible for development of a CMML-like phenotype is not yet known. However, several candidate genes have been identified and manipulated in mouse models and shown to develop a CMML-like phenotype. Alterations in the tumor suppressor gene, Notch1, were studied by knocking down Nicastrin (Notch-derived signal) and shown to aberrantly skew hematopoiesis towards a myeloid compartment bias, thus producing a CMML-like phenotype with monocytosis, myeloproliferation, bone marrow dysplasia and progression to AML²⁵. However, specific inactivating Notch1 pathway mutations have not shown to be frequently altered in CMML²⁵. More recently, several independent groups have shown that concurrent TET/NRAS^G12D alterations in hematopoietic cells induce an aggressive CMML-phenotype by disrupting the mitogen-activated protein pathway^26,27. These studies conclusively establish the importance of epigenetic and RAS pathway perturbations in developing CMML phenotype, which thereby provide opportunities for therapeutic modulation. Alternative pathways such as autophagy have also shown to be important and under investigation²⁸.

III. Immune tolerance mechanisms:

Given the advent of immunotherapy and other cell-based drugs, the role of putative genomic abnormalities in inducing pathways that contribute towards local and systemic immune tolerance in CMML is another key area of active investigation. In approximately 20% CMML patients, autoimmune diseases and/or systemic inflammatory syndromes exist concurrently²⁹, however whether this is a cause or consequence of the disease is still unclear. A study looking at expression of known immune checkpoints confirmed an increased relative programmed death-ligand (PD-L1) expression in CMML, compared to AML and its upregulation upon treatment with hypomethylating agents (HMA), thus indicating a possible therapy resistance mechanism³⁰. Additionally, role of myeloid-derived suppressive cells has been suggested to inhibit innate and acquired immune responses^31,32. Dendritic cells nodules comprising of mainly plasmacytoid dendritic cells have long been known to coexist in the bone marrow microenvironment of approximately 20% CMML patients^33–35. In the appropriate biologic context, they can either be immunostimulatory or immunosuppressive³⁶, however their specific role in CMML has not yet been explored. These findings clearly support the role for local and systemic immune suppression in contributing towards disease biology of CMML and are worth investigating.

Diagnosis:

Diagnosis for CMML is suspected in a patient with persistent monocytosis. An approach for the work-up of other malignant and benign causes of monocytosis is outlined in ref.⁴ As discussed above, the 2016 WHO criteria for establishment of diagnosis of CMML needs persistent (>3 months) PB monocytosis (≥ 1 × 10⁹/L; ≥ 10% of WBC differential); absence of WHO-defined criteria for diagnosis of BCR/ABL1-driven chronic myeloid leukemia, and JAK2/MPL/CALR-driven polycythemia vera, essential thrombocythemia and polycythemia vera; absence of disease-defining gene rearrangements/fusions such as PDGFRA, PDGFRB, FGF1 and PCM1-JAK2; <20% blasts or blast equivalents (monoblasts, myeloblasts and promonocytes) in both PB and bone marrow (BM), and dysplasia in one or more myeloid cell lineages [if the degree of dysplasia is insufficient, diagnosis can still be established in the presence of somatic variants such as TET2, SRSF2, ASXL1 and/or SETBP1².

Over the last few years, the role of flow cytometry in distinguishing monocyte subsets [Classical (MO1; CD14⁺/CD16⁻); intermediate (MO2; CD14⁺/CD16⁺) and nonclassical (MO3; CD14⁻/CD16⁺)] has contributed towards the diagnosis of CMML^37,38, and its distinction from other reactive and non-CMML malignant causes of monocytosis; CMML patients demonstrating a characteristic increase in the classical monocyte fraction (>94% of total monocytes)³⁸. The associated specificity and sensitivity of this approach were found to be 95.1% and 90.6% respectively³⁸, however false negatives due to increase in MO2 fractions in cases with co-existing autoimmune conditions need to considered³⁹. Prospective evaluation of other cell surface markers such as CCR2, CD11c, CD36 and HLA-DR⁴⁰, and novel technologies such as mass cytometry⁴¹ will not only help enhance the diagnostic accuracy of this technology, but also potentially aid in its development of useful biomarkers for predicting treatment response.

Bone marrow morphology and immunohistochemistry evaluation is part of the standard work-up for diagnosis of CMML. Although no single morphologic feature is diagnostic, granulocytic hypercellularity and dysplasia are frequent findings. Immature and mature monocytes with precursors such as promonocytes/monoblasts are also often seen, and promonocytes and monoblasts are considered as blast equivalents¹. Presence of dysplasia, however, is not a requirement for CMML diagnosis if pathogenic alterations in any of the frequently altered genes, such as TET2, ASXL1, SRSF2, and SETBP1 are demonstrated. However, given the fact that these gene mutations can present as age-related mutant clones (CHIP), genetic findings must always be cautiously interpreted in relation to appropriate clinical associations. Along with morphologic assessment, cytochemistry studies using alpha-naphthyl butyrate esterase or alternatively, alpha-naphthyl acetate esterase (with fluoride inhibition) staining is recommended, either in isolation or in combination with naphthol AS-D chloroacetate esterase staining to differentiate monocytes from monoblasts and promonocytes and from non-monocytic cells^1,2,4.

Immunophenotypic markers most commonly known to characterize CMML are classic myelomonocytic antigens (CD13, CD33), along with differential expression of CD14, CD64 and CD68. The most reliable marker for identification of monocytic cells is CD14, while for identification of associated mature plasmacytoid dendritic cells (pDC) islands, CD123 is most commonly used at our institution, although other markers such as CD303, CD2AP, CD45RA, BCL11A and TCL-1 have also been shown to identify the same¹.

Classification:

The 2016 revision to the WHO classification includes several changes to morphology based CMML subtypes^1,2. A three-tier blast-based system replaced the original two-tier system, with ‘CMML-0’ category introduced with the following sub-categories: CMML-0 (PB <2% and/or BM <5%), CMML-1 (PB 2–4% and/or BM 5–9%), and CMML-2 (PB 5–19% and/or BM 10–19% and/or presence of Auer rods. Historically, the French-American-British (FAB) CMML classification included 2 subtypes, based on white blood cell (WBC) count cut-off of 13 × 109/L: myelodysplastic-CMML (MDS-CMML; WBC count <13 × 109/L) and myeloproliferative CMML (MP-CMML; WBC count ≥ 13 × 109/L). These revisions have been shown to be prognostically relevant in a single-institution cohort⁴². Although sub-classification based on WBC count may seem arbitrary, these two subtypes remain genomically distinct, with the MP-CMML subtype enriched for RAS pathway mutations⁴³. With that being said, due to transient differences of WBC count, some patients still float between the two categories, thus calling for a more stable, genomically derived classification. Given the clinical heterogeneity, such a classification would allow development of clinical trial protocols for tailored therapies and individualized care.

Prognosis:

International Prognostic Scoring System (IPSS) and its subsequent revision (IPSS-R) were primarily designed for MDS patients and excluded patients with the myeloproliferative subtype of CMML (MP-CMML)^44,45. Thereafter, several CMML-specific models were published; however, only four (CPSS, GFM, Mayo Prognostic and CPSS-Molecular) models have been validated, which we have discussed in this review. For details on other models, we refer the reader to ref.⁴

In 2013, Such and colleagues developed the CMML-specific prognostic Scoring System (CPSS) in a Spanish cohort (training) of 558 patients, and validated it using an independent cohort of 274 patients from centers in Germany and Italy⁵. This model was able to separate patients into 4 risk groups (stratified by both OS and risk for transformation into AML), based on FAB and WHO CMML subtypes, CMML-specific cytogenetic risk stratification⁹ and RBC transfusion dependency. Each variable garnered 1 point, except high-risk cytogenetics which earned 2 points, and the four risk categories were: low (0 points, median OS: 72 months), intermediate-1 (1 point, median OS: 31 months), intermediate-2 (2–3 points, median OS: 13 months), and high (4–5 points, median OS: 5 months)⁴⁶.

Later in 2013, the Mayo Prognostic Model was developed and reported in 226 CMML patients evaluated in Rochester (Minnesota), and validated in an independent cohort of 268 patients from Tampa (Florida)⁴⁷. Increased absolute monocyte count (AMC, >10 × 10⁹/L), presence of circulating immature myeloid cells, decreased hemoglobin (<10 g/dL), and decreased platelet count (<100 × 10⁹/L) were identified as risk factors with an independent prognostic impact. Variant analysis was done for genes; ASXL1, SF3B1 and SRSF2, with Sanger sequencing and standard polymerase chain reaction/bidirectional sequencing, however pathogenic alterations in neither of three genes retained an independent prognostic impact in a multivariable model. Risk stratification included the following sub-categories: low (0 risk factors, median OS: 32 months), intermediate (1 risk factor, median OS: 18.5 months) and high risk (2 or more risk factors, median OS: 10 months)⁴⁷.

Subsequently in 2013, the French group published the Groupe Francophone des Myelodysplasies (GFM) model⁴⁸, incorporating gene mutations for the first time, in a CMML-specific prognostic model. An independent prognostic impact was determined by ASXL1 mutational status (yes: 2 points), age ≥ 65 years (1 point), WBC count ≥ 15 × 10⁹/L (3 points), platelet count ˂ 100 × 10⁹/L (2 points), and anemia [(hemoglobin < 10 g/dL in females, <11 g/dL in males), yes: 2 point). The three prognostic categories were ascertained as low (0–4 points, median OS not reached), intermediate (5–7 points, median OS 38.5 months), and high (8–12 months, median OS 14.4 months)⁴⁸. This model was validated with both an internal dataset with bootstrapping analyses and in an external German cohort⁴⁸. In the Mayo model, all frameshift, missense and nonsense mutations in the ASXL1 gene were included, while in the GFM model, only frameshift and nonsense mutations were included, which probably explained why an independent prognostic was ascertained in the latter but not in the former model. Subsequently, when only frameshift and nonsense ASXL1 mutations were included in a combined Mayo-French cohort (Mayo Molecular Model), they retained an independent prognostic impact, along with AMC > 10 × 10⁹/L, Hb <10 gm/dL, circulating IMC and platelet count <100 × 10⁹/L¹². Risk categorization in this model included low (no risk factors, median OS: 16 months), intermediate-1 [one risk factor, HR 1.9 (95% CI 1.1–3.3), median OS: 31 months], intermediate-2 [2 risk factors, HR 3.4 (95% CI 2–5.6), median OS: 59 months], and high risk categories [≥ 3 risk factors, HR 6.2 (95% CI 3.7–10.4), median OS: 16 months]¹². Although this model included patients from two different centers, validation was not done in an independent cohort.

Recently, the CPPS-molecular model included a combination of clinical/genetic variables; genetic risk groups (HR 1.7, p<0.001), RBC transfusion dependency (HR 2.5, p<0.001), WBC count (HR 2.3, p=0.001), BM blasts (≤ 5%, HR 1.6, p=0.04). Four risk groups were identified using regression analysis; low (zero risk factors, median OS not reached), intermediate-1 (1 risk factor), intermediate-2 (2–3 risk factors) and high-risk (≥4 risk factors) groups⁴⁹. The four-year leukemic transformations were found to be 0%, 3%, 21% and 48% respectively⁴⁹. Table 2 provides a summary and details of the four validated prognostic models.

Table 2:

Table providing details of four validated CMML-specific prognostic models.

Model	Training cohort	Validation cohort	Variables included	Risk categories	Overall survival	Risk of transformation to AML
CPSS	578	274 (external)	CMML FAB subtype, CMML WHO subtype, CMML-specific cytogenetics, RBC transfusion dependence			Risk of AML evolution at 5 years:
				Low	72	13%
				Intermediate-1	31	29%
				Intermediate-2	13	60%
				High	5	73%
Mayo Prognostic Model	226	268 (external)	Increased AMC > 10 × 109/L, presence of circulating blasts, hemoglobin < 10 gm/dL, platelet count < 100 × 109/L	Low	32	NR
				Intermediate	18.5
				High	10
GFM model	312	165 (external)	Age > 65 years, WBC > 15 × 109/L, anemia, platelets < 100 × 109/L, ASXL1 mutation			AML-free survival (in months):
				Low	Median not reached	56
				Intermediate	38.5	27.4
				High	14.4	9.2
CPSS-Molecular	214	260 (external)	Genetic risk groups defined by CPSS cytogenetic risk stratification and pathogenic gene variations in ASXL1, NRAS, SETBP1 and RUNX1 genes; BM blasts > 5%, WBC count > 13 × 109/L, red blood cell transfusion dependence			48 months cumulative incidence of AML:
				Low	Median not reached	0%
				Intermediate-1	64	3%
				Intermediate-2	37	21%
				High	18	48%

Abbreviations: CMML = chronic myelomonocytic leukemia; CPSS = CMML-specific prognostic scoring system; GFM = Groupe Francophone des Myélodysplasies; FAB = French-American-British; WHO = World Health Organization; BM = bone marrow; AML = acute myeloid leukemia.

Future developments in refining the existing prognostic classifications would be to look beyond genomic signatures and incorporate other biologically relevant prognostic factors such cytokine signatures²¹ and methylation changes. Building a self-learning analytical model through machine learning algorithms and incorporating genomics, epigenetics, cytokine profiles and relevant clinical variables would enable personalized prediction of relevant clinical events for an individual patient in the clinic, a feat currently not possible with traditional prognostic models. However, achieving this would need a collaborative effort and creation of a large, uniform international patient database.

Therapy:

As iterated previously, CMML primarily affects elderly individuals, with a median age at diagnosis ~70–75 years, and therefore, most patients are ineligible for aggressive therapeutic interventions. Limited treatment options exist for CMML, with only allogeneic hematopoietic stem cell transplantation (HSCT) capable of providing a cure, albeit with substantial treatment-related morbidity and mortality. As of yet, there are only two drugs approved by the United States Food and Drug administration, namely 5-azacitidine and decitabine. In 2002, the first phase 3 randomized control trial of 5-azacitidine in patients with MDS was reported by the Cancer and Leukemia Group B⁵⁰. One hundred-ninety one patients were randomized to receive either 5-azacitidine (n=99, 52%) at 75 mg/m²/day for 7 days every 28 days versus best supportive care (BSC, n=92, 48%). Of them, 7 patients with a diagnosis of CMML were randomized to both arms each. Although the overall response rate (ORR) in the 5-azacitidine arm was 60% versus 47% in BSC, CMML-specific response rates were not reported. Transformations into AML as a first event was lower (15 versus 30%) with an improved median OS (20 versus 14 months) in favor of 5-azacitidine. Significant grade 3 or higher toxicities in the 5-azacitidine arm were mainly hematologic; leukopenia (59%) and thrombocytopenia (70%)⁵⁰. In 2006, a randomized phase 3 trial was published which included 170 MDS (14 CMML) patients, randomized to decitabine at 15 mg/m² for 3 days every 6 weeks or best supportive care arms⁵¹. Although, the median OS was not significantly different in the two groups, time to transformation into AML was higher (12.1 months versus 7.8 months) in the decitabine arm⁵¹. The beneficial effects of 5-azacitidine over best supportive care was confirmed in another pivotal phase 3 trial by Fenaux P and colleagues of 358 high-risk MDS (16 CMML) patients, showing an improvement in OS (24.5 versus 15 months) and increased median time to transformation into AML (15 versus 10.1 months) in favor of the 5-azacitidine arm⁵². A significant limitation of these trials is the fact that they were primarily designed for and included MDS patients and excluded CMML patients with proliferative features. Further, a study from Merlevede and colleagues, assessing comprehensive mutational landscape in serial CMML samples, showed that HMAs do not decrease the mutational allele burden in CMML, thus acting predominantly through an epigenetic mechanism⁵³. In addition, their results highlighted that response to a HMA does not prevent the accumulation of genetic damage responsible for emergence of an AML subclone, as indicated in a patient who had a partial response to decitabine for 2 years until an EZH2/ETV6 mutated subclone emerged resulting in an AML phenotype. Although cytotoxic effect cannot be completely excluded with these results, immunophenotyping studies suggested that these drugs cannot eradicate leukemic stem or progenitor cells and are limited to reversing the epigenetic alterations. Hence, HMAs are limited in their effect on disease biology with a limited potential for cure, despite being useful palliative agents.

Other clinical investigational agents such ruxolitinib (JAK 1/2 inhibitor)⁵⁴, SL-401 (anti-CD123/IL-3Rα agent), H3B-8800 (small molecule splicing modulator) and lenzilumab (anti-GM CSF antibody), are at various stages of clinical trial investigation⁴. Combined results from a phase 1 and 2 study assessing ruxolitinib in 49 CMML patients (56% with MP-CMML) reported an overall response rate of 46%, as defined by the MDS/MPN International Working Group criteria⁵⁵, along with improvement in symptoms as assessed by the MPN symptom assessment form. Median OS was found to be prolonged in the treatment cohort, compared to a historical cohort (69 versus 31 months, p=0.03), however results need to be interpreted with caution due to the inherent limitations of such an analysis and need confirmation in a large, randomized phase 3 study⁵⁶. SL-401, a novel targeted drug directed against interleukin-3 receptor (CD123), was studied in a phase 1/2 trial of a limited cohort of 14 relapsed/refractory CMML patients, and demonstrated 71% (5/7) spleen reductions and 17% (2/12) rate of bone marrow complete response (responses were durable with 1 complete response for 15+ months, while 1 bone marrow complete response for 4+ months)⁵⁷.

Similar to MDS, other palliative interventions include hydroxyurea for patients with proliferative CMML, use of erythropoietin analogues, prophylactic antibiotics for persistent neutropenia, splenectomy for persistent splenomegaly-related symptoms⁵⁸ and iron chelators for patients dependent on RBC transfusions⁴. No CMML-specific guidelines exist on how to utilize these palliative treatments and should be recommended for patients at the discretion of the treating hematologist.

Allogeneic HSCT remains the sole curative option for patients with CMML, however the optimal timing and patient selection remains a matter of debate. Retrospective data, mostly from single-institution studies, indicates comparable rates (~12–50%) of response rates and treatment-related mortality^4,59–65. A critical question for clinicians is whether allogeneic HSCT overcomes the adverse prognostic impact of patients with higher risk CMML subtypes, however prospective data to address this question is lacking. At our institution, we recommend upfront allogeneic HSCT if CMML is diagnosed with intermediate or high risk disease in younger patients, especially if they have a good performance status (ECOG ≤ 2), low co-morbidity index and matched sibling donors. Transplantation-specific prognostic models are necessary to predict likelihood of disease relapse and transplant-related morbidity and mortality. A single-institution study found advanced age, increased HCT comorbidity index and adverse-risk cytogenetics to have an independent adverse prognostic impact with greater mortality and decreased relapse-free survival⁵⁹. Retrospective data from Center for International Blood and Marrow Transplant Research validated the CPSS-prognostic system in 209 CMML patients who were treated with HSCT, and found Karnofsky performance status, graft source and CPSS score to be independent predictors of survival in a multivariable model⁶⁶. Recently in MDS, mutation clearance detected in the bone marrows 30 days after allogeneic HSCT, correlated with a lower risk of disease progression⁶⁷. Whether the same conclusions apply to CMML is currently unknown. In 2017, an international panel of experts published consensus guidelines on allogeneic HSCT for CMML patients. In making treatment decisions, CPSS was the preferred model; however IPSS-R could also be used for dysplastic CMML patients⁶⁸. Achievement of remission prior to HSCT was the most important prognostic factor of a favorable outcome, and upfront transplant was recommended for patients with CPSS intermediate-2 or high risk disease. However, prospective clinical trial validation is lacking in this regard⁶⁸.

Conclusion:

In summary, there have been several advances in CMML in terms of disease prognostication and investigation into targeted therapeutic approaches, owing to the advent of precision genomics. Despite this, we do not have effective disease-modifying agents with a distinct survival benefit. Novel sequencing technologies have enabled identification of putative molecular drivers for the disease and thus provide a benchmark to base future classification systems. Although targeted therapy bears the promise of individualizing care for patients, there is still limited curative potential. Immune and other cell-based therapies remain largely unexplored in this disease, thus providing opportunities for future novel disease-modifying therapies.

Table 3:

Table summarizing major completed and ongoing clinical trials for patients with chronic myelomonocytic leukemia

Drug (Year trial published)	Mechanism of action	Clinical trial	Number of CMML patients included	Approval by the US FDA	Response rate	Grade ≥ 3 toxicity in treatment arm	Transformation into AML	OS
Azacitidine (2009)	Hypomethylating agent	Phase 3	16 (358 in total)	Approved	CMML-specific response rate NR	Neutropenia (91%) Thrombocytopenia (85%), Anemia (57%), deaths within first 3 months (11%).	Median time to AML transformation: 15 months in Azacytidine arm versus 10.1 months in BSC arm.	At median follow-up of 21.1 months for the entire cohort: 24.5 months (Azacytidine) versus 15 months (conventional care)
Decitabine (2006)	Hypomethylating agent	Phase 3	14 (170 in total)	Approved	1/6 ITT (17%)	Neutropenia (93%), thrombocytopenia (17%), anemia (15%), febrile neutropenia (28%), leukopenia (17%), pyrexia (7%), hyperbilirubinemia (7%), pneumonia (18%), nausea (1%), constipation (2%), abdominal pain (2%), oral mucosal petechiae (2%)	Time to AML transformation: 12.1 months in decitabine arm versus 7.8 months in BSC arm (p=0.16, log-rank test).	OS 14 months in decitabine arm versus 14.9 months in BSC arm (p=0.636). OS in decitabine responders 23.5 months versus 13.7 months in non-responders (p=0.007).
Azacytidine (2002)	Hypomethylating agent	Phase 3	14 (191 in total)	Approved	CMML-specific response rate NR	Leukopenia (59%; granulocytopenia 81%), thrombocytopenia 70%	Transformation to AML as first event: 15% in azacytidine arm versus 30% in BSC.	Median OS in treatment arm was 20 months versus 14 months undergoing BSC
Azacitidine plus lenalidomide, and Azacitidine plus vorinostat (2017)	Hypomethylating agent plus immunodulation (lenalidomide) or HDAC inhibition (vorinostat).	Randomized phase 2	53	Not approved yet	ORR: 38% (49% in Azacitidine plus lenalidomide, p=0.14; 27% in Azacitidine plus vorinostat, p=0.16)	Azacitidine plus lenalidomide: Cytopenias 16 (18%), infections 14 (16), skin 14 (16%), maximum grade any adverse effect 59 (66%) Azacitidine plus vorinostat: Cytopenias 13 (14%), GI 14 (15%), maximum grade any adverse effect 52 (57%).	NR	Median not reached in all groups
Ruxolitinib (2017; abstract)	JAK inhibitor	Phase 2	29 (56% proliferative CMML)	Not approved yet	ORR 5 (74%)	Toxicities similar to phase 1 (details not available)	NR	Median OS 69 months in ruxolitinib arm compared to 31 months (historical cohort)
Lenzilumab (not reported)	Anti-GM CSF monoclonal antibody	Phase 1	Accruing	Not approved yet	NA	NA	NA	NA
Tipifarnib	Farnesyl transferase inhibitor	Phase 2	Accruing	Not approved yet	NA	NA	NA	NA
SL-401	Anti-interleukin 3 receptor (CD123)	Phase 1/2	Accruing	Not approved yet	NA	NA	NA	NA

Abbreviations: CMML = chronic myelomonocytic leukemia; AML = acute myeloid leukemia; OS = overall survival; ITT = intention-to-treat; BSC = best supportive care; HDAC = histone deacetylase; ORR = overall response rate; NR = not reached; GM-CSF = granulocyte monocyte colony stimulating factor.