RAS Mutations Contribute to Evolution of Chronic Myelomonocytic Leukemia to the Proliferative Variant

Chronic myelomonocytic leukemia is a rare malignancy of the elderly whose diagnosis entails persistent absolute monocytosis in the blood (monocytes, >1 × 10⁹/L). It exhibits highly heterogeneous clinical, hematologic, and morphologic features, varying from mainly myelodysplastic to predominantly myeloproliferative (1, 2). Hence, chronic myelomonocytic leukemia nosology has always represented a hurdle; the French American Group (FAB) classification firstly categorized this disorder as a myelodysplastic syndrome (3) and then, in 1994, based on the WBC count threshold of 13 × 10⁹/L, proposed to discriminate between myelodysplastic chronic myelomonocytic leukemia (MD-CMML, WBC, <13 × 10⁹/L) and myeloproliferative chronic myelomonocytic leukemia (MP-CMML, WBC, ≥13 × 10⁹/L; ref. 4). Yet, the biological and clinical significance of this distinction remains controversial because the two groups have been variably associated with prognosis (5–10) and few cases of MD-CMML progressing to MP-CMML have been reported (1, 2, 8, 11); accordingly, this discrimination was not acknowledged in the last WHO classification of myeloid malignancies, in which chronic myelomonocytic leukemia has been rather included in the new category of the myelodysplastic syndrome/myeloproliferative neoplasms (12, 13).

The high variability of the clinical presentation and course of chronic myelomonocytic leukemia reflects the heterogeneity of its pathogenetic features. Since, to date, the recurrent mechanisms involved in its malignant evolution have not been fully elucidated, chronic myelomonocytic leukemia remains a diagnostic and therapeutic challenge, in most cases with a dismal prognosis. Notably, among hematologic malignancies, chronic myelomonocytic leukemia encounters the highest incidence of RAS substitutions (up to 50%, usually in codons 12, 13, and 61 of NRAS and KRAS; ref. 14), which have been associated with features of cell proliferation and monocytosis (1) and with shorter survival (15). Of note, a significantly higher incidence of these mutations has been documented among MP-CMML than MD-CMML patients (1, 14), and the expression of activated KRAS and NRAS in murine bone marrow cells has been shown to rapidly and efficiently induce a fatal myeloproliferative disease resembling human chronic myelomonocytic leukemia (16, 17). However, RAS mutations are not specific for chronic myelomonocytic leukemia (18), and conversely, other genetic lesions have been found in subsets of patients. For example, RUNX1 aberrations have been detected in MD-CMML and MP-CMML patients, also in concomitance with RAS mutations, suggesting implication of these abnormalities in the acquisition of myelodysplastic features (14). More recently, mutations of the putative tumor suppressor gene TET2 have been identified in a spectrum of myeloid malignancies (19–26), including chronic myelomonocytic leukemia (21–23, 26), MD-CMML and MP-CMML (26). Noteworthy, the ubiquitous nature of TET2 mutations suggests that they are an early event in hematopoietic differentiation and that at least one additional contributing event is required to determine the disease phenotype (19, 27). Other genetic defects (i.e., FLT3-ITD and JAK2 V617F substitution; refs. 28–30) have been identified at a higher frequency among MP-CMML than MD-CMML patients, in a manner mutually exclusive with RAS mutations, and have been shown to induce a chronic myelomonocytic leukemia-like disease in mice (i.e., the FLT3-ITD and BID deletion; refs. 28, 31), implying a role in the establishment of myeloproliferation. Therefore, although a key role for RAS mutations in chronic myelomonocytic leukemia pathogenesis has been proposed, particularly in those forms with a prominent proliferative component, no straightforward conclusion has been definitely drawn.

Because its pathogenesis is mostly unknown, to date, the only possible curative strategy for chronic myelomonocytic leukemia remains allogeneic stem cell transplantation, which, however, may be applied only to a small fraction of patients because of the advanced median age of presentation (32). Hence, to improve patient outcome, it is crucial to define molecular subtypes of chronic myelomonocytic leukemia, which could possibly facilitate the development of novel targeted agents.

In light of these issues, we aimed to investigate the biological and clinical significance of the discrimination between MD-CMML and MP-CMML. Our findings support the prognostic significance of this distinction because MP-CMML patients had a significantly shorter survival than those with myelodysplastic disease. Moreover, we document for the first time the expansion of a RAS-mutated clone in concomitance with progression from MD-CMML to MP-CMML, strongly suggesting these aberrations to function as a secondary event that contributes to the expansion of a malignant clone with proliferative advantage. Altogether, our results highlight the clinical relevance of identifying chronic myelomonocytic leukemia patients harboring these aberrations because they are predicted to develop the disease phenotype with the worse prognosis (MP-CMML).

Forty patients diagnosed with chronic myelomonocytic leukemia were included in this study between December 2002 and December 2008; according to the FAB proposal (4), 22 were classified as MD-CMML and 18 as MP-CMML. In addition, our series included one patient (patient 38) who developed chronic myelomonocytic leukemia rapidly progressing to the MP-CMML, after achievement of complete remission from a previously diagnosed Philadelphia-positive chronic myelogenous leukemia (CML). Clinical and laboratory characteristics of patients are listed in Tables 1 and 2. Five patients (patients 7, 8, 9, 31, and 33) initially presented as MD-CMML and later on experienced evolution to myeloproliferative disease. Patients 24 and 30 were referred as MP-CMML but had a previous history of MD-CMML.

Overall, 62 peripheral blood samples were obtained after informed consent. Peripheral blood mononuclear cells were isolated by Hystopaque 1077 density gradient (Sigma). Genomic DNA extraction was carried out by DNAzol (Molecular Research Center, Inc.); RNA was isolated by TRIreagent (Sigma), and cDNA was synthesized using oligo(dT)_12-18 primers and SuperScript II Reverse Transcriptase (Invitrogen) according to the manufacturer's instructions.

Regions spanning codons 12 to 13 (exon 1) and 61 (exon 2) of NRAS and codons 12-13 (exon 1), 61 (exon 2), and 146 (exon 3; refs. 14, 33) of KRAS were amplified by PCR using M13-tailed primers starting from 200 ng of DNA (Supplementary Table S1 and S2). Amplicons were purified (MinElute QIAquick gel extraction kit, Qiagen) and automatically sequenced (ABI PRISM 310 Genetic Analyzer, Applied Biosystems) using the Big Dye Terminator Cycle Sequencing Kit.

Primers for amplification of the entire human NRAS coding sequence (GenBank NM_002524) were designed as previously described (34), with minor modifications. Briefly, two nucleotides were added to the forward primer to insert NRAS in frame with GFP, and four and two excess nucleotides were added at the 5′ of PstI and BamHI sites, respectively, to increase cleavage efficiency of endonucleases (Supplementary Table S1). The sequence of the final construct was verified by sequencing using the internal primer (NRAS coding sequence-forward internal) specified in Supplementary Table S1. PstI/BamHI-digested amplicons corresponding to the entire coding sequence of NRAS wild type (WT), G12D, and G60E were inserted into the PstI/BamHI-digested pAcGFP-C1 vector (Clontech Laboratories, Inc.), leading to a chimeric construct of GFP fused in frame to the 5′ end of NRAS. The resulting plasmids were subcloned into Escherichia coli.

NIH3T3 mouse fibroblasts were maintained in DMEM (Cambrex Corporation) supplemented with 10% of fetal bovine serum (Sigma) and 1% of a penicillin/streptomycin solution (Sigma). Cells were transiently transfected with 1 μg of the pAcGFP-C1 vector, either empty or containing the GFP-NRAS WT, G12D, or G60E constructs, using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.

For confocal analysis, NIH3T3 cells were grown on four-well glass chambers, transfected, fixed 48 h later with 4% paraformaldehyde, permeabilized with 1% Triton X-100, and incubated overnight with 1:50 Anti-Pan Ras Ab (Calbiochem). After 30 min of labeling with phycoerythrin, cells were washed, mounted with cover glass, and analyzed by confocal microscopy.

For biochemical analysis, 48 h after transfection, NIH3T3 cells were lysed to pull down the GTP-bound Ras by affinity purification with the Raf-1RBD agarose (Pierce Biotechnology, Inc.) following the manufacturer's instructions. Whole cell extracts and pull-down supernatants were then separated on SDS-PAGE, and immunoblots were incubated overnight with the Anti-Pan Ras Ab (Calbiochem).

Either forward or reverse primers were designed with specific mismatches in codons 12, 13, and 60 of NRAS and in codon 12 of KRAS (Supplementary Table S1). To improve specificity, an intentional mismatch on the third nucleotide from the 3′ end was also included. Addition of the M13-tailed primers to the reaction led to the amplification of either the internal control fragment alone (RAS WT) or concomitantly with the specific mutant allele (Supplementary Tables S1 and S2). One hundred nanograms of DNA were used in a final volume of 25 μL. Allele-specific PCR specificity was assessed by amplifying multiple DNA from healthy donors (RAS WT) and from patients harboring mutant (mut) RAS alleles other than that specific for each allele-specific PCR. To determine allele-specific PCR sensitivity, different amounts of mutDNA were spiked into DNA WT and subjected to the corresponding allele-specific PCR (except for the NRAS G13V mutation because of the paucity of starting material).

Peripheral blood mononuclear cells were mixed with a methylcellulose culture medium (MethoCult 4434, StemCell Technologies, Inc.) and single colony–forming unit, granulocytes-macrophages (CFU-GM), and burst-forming unit-erythroid (BFU-E) were harvested in 40 μL of deionized water and heated for 10 min at 95°C to release DNA. Allele-specific PCR for the presence of the mutation previously identified in the total DNA was done.

For the survival analyses, Kaplan-Meier plots were generated using the elapsed time between the date of the blood collection and death or the date of the last follow-up, and the survival curves were compared using log-rank testing. Unpaired one-tailed t test was used to test the difference for continuous variables. The Fisher's exact test was used to examine the association of two categorical variables. P < 0.05 was considered statistically significant. The level of agreement between allele-specific PCR and sequencing was assessed using the κ statistics, considering substantial agreement if κ was >0.61. Analyses were carried out using GraphPad Prism version 4.00 for Windows (GraphPad Software; www.graphpad.com).

The entire cohort of chronic myelomonocytic leukemia patients included 23 men and 17 women, with a median age of 74 years (range, 59-89 years). The median follow-up was 28.3 months (range, 0-82.5 months), and the median survival was 40 months (Fig. 1A). When patients were classified according to the WBC count into MD-CMML (n = 22) and MP-CMML (n = 18), the median WBC counts resulted 6.3 × 10⁹/L and 15.8 × 10⁹/L, respectively. The median age was 73 years for MD-CMML and 75 years for MP-CMML patients. As predicted, the two groups significantly differed for the absolute monocyte count (median, 1.8 × 10⁹/L in MD-CMML versus 3.7 × 10⁹/L in MP-CMML; P < 0.0001). Importantly, according to the Kaplan-Meier estimates, patients with MP-CMML had a significantly shorter median survival than those with MD-CMML (14.5 months versus not reached; P = 0.003; Fig. 1B).

We directly sequenced exons 1 and 2 of NRAS and 1 to 3 of KRAS. Overall, at first presentation, no mutation was detected in MD-CMML patients, whereas 4 (22.2%) of 18 MP-CMML patients were shown to harbor a mutant allele (P = 0.033), the most frequent being the NRAS G12D (Table 2). Notably, patient 30, who carried the KRAS G12R mutation, was referred to our institution as MP-CMML but had a previous history of MD-CMML; unfortunately, no sample was available at the time of myelodysplastic disease.

During follow-up, sequencing identified the NRAS G60E and G12D substitutions in two patients (patients 8 and 33, initially presenting as MD-CMML with RAS WT) after they progressed to MP-CMML. With regard to the MP-CMML group, sequencing detected the NRAS G12V mutation in patient 24 36 months after the time of presentation and the NRAS G12R and G13V alleles in patient 38 not at the time of presentation as Philadelphia-positive CML but at different times during the course of chronic myelomonocytic leukemia.

Because of the lack of data on the uncommon NRAS G60E mutation at the time of its identification in patient 8 and to better understand its role in chronic myelomonocytic leukemia evolution, we expressed the GFP/NRAS WT, G60E, and G12D proteins in NIH3T3 cells and compared their biochemical activity. After verifying by confocal microscopy the expression of chimeric proteins in transiently transfected cells (Supplementary Fig. S1A), we determined the levels of active GTP-bound Ras. As shown in Supplementary Fig. S1B, increased levels of Ras-GTP were observed in cells expressing the mutant proteins compared with the WT; in particular, the NRAS G60E displayed intermediate levels of active Ras between the WT and the G12D, in agreement with the recent paper from Tyner et al. (33).

To gain further insights into the role of activated RAS in chronic myelomonocytic leukemia evolution, we set up an allele-specific PCR for each substitution identified by sequencing and rescreened all specimens. In our experimental conditions, allele-specific PCR was able to detect the presence of the corresponding mutant allele in a mixture containing ∼1% to 2% of mutDNA in a background of DNA WT (Supplementary Fig. S2), whereas sequencing required about 20% (results not shown).

Allele-specific PCR confirmed all results obtained by sequencing. Yet, this method led to some interesting findings summarized in Table 2 (the κ statistics indicated a substantial agreement between allele-specific PCR and sequencing; κ = 0.6154). Overall, at presentation, a RAS mutation was detected in 2 (9%) of 22 MD-CMML and 6 (33.3%) of 18 MP-CMML patients (P = 0.065). Importantly, both MD-CMML patients presenting with a mutant allele (patients 8 and 33) later on progressed to MP-CMML; in our experimental conditions, a small amount of the NRAS G60E allele was detected in patient 8's peripheral blood withdrawn 22 months before disease progression (Fig. 2A), and the NRAS G12D allele was found at low levels in the peripheral blood obtained from patient 33 at the time of first presentation (Fig. 2B). In this case, three more samples were consecutively collected 13, 18, and 25 months after the first withdrawal; of note, allele-specific PCR allowed to monitor the increase of the level of the oncogenic allele throughout the course of chronic myelomonocytic leukemia (Fig. 2B and Supplementary Fig. S3). Of the other three patients who experienced evolution to myeloproliferative disease, patients 7 and 9 did not display any of the alterations detectable by our assays, whereas patient 31 was shown to harbor the FLT3-ITD (results not shown).

Among MP-CMML patients, at presentation, allele-specific PCR identified low levels of the NRAS G12D allele in patients 22 and 32, who were both WT according to sequencing. During the follow-up, allele-specific PCR could detect the NRAS G12V allele in patient 24 27 months after the first presentation, 9 months earlier than sequencing (Fig. 2C). In patient 38, who developed chronic myelomonocytic leukemia as a second malignancy after remission from Philadelphia-positive CML, in our experimental conditions, no mutation was found at the time of CML (Fig. 2D). Later on, allele-specific PCR allowed to accurately follow the consecutive occurrence of two distinct clones harboring the NRAS G13V and G12R alleles, respectively. In particular, allele-specific PCR identified the presence of the G12R allele ∼30 months earlier than sequencing and the persistence of the G13V mutation 17 months after sequencing could no longer detect it.

Finally, single colony analysis by allele-specific PCR was carried out to assess in which progenitors the G12D mutation occurred. We plated peripheral blood mononuclear cells from patient 12 (MP-CMML with mutRAS by sequencing) and from patients 32 and 33 (respectively, MP-CMML and MD-CMML at first presentation, with RAS WT by sequencing) and found the mutation in colony-forming unit, granulocytes-macrophage, and BFU-E, exception made for patient 32, in which case no BFU-E grew. Interestingly, as shown in Table 3, despite the relatively low number of colonies analyzed, when comparing patients 12 and 33 at time 0 the percentage of cells carrying the G12D allele reflected the level of mutation found by allele-specific PCR (Fig. 2B).

We analyzed the impact of RAS mutations on survival of chronic myelomonocytic leukemia patients. The median age was 74 years in WT and mutated groups. The group of patients presenting with mutRAS by allele-specific PCR (n = 8) had a significantly higher WBC count compared with the group with RAS WT (n = 32; median, 13.6 × 10⁹/L versus 9.7 × 10⁹/L, respectively; P = 0.035). Importantly, although median survivals of patients with mutRAS and RAS WT were not significantly different, possibly because of the limited sample size, a trend for a shorter life expectation, more evident after the first 2 years of follow-up, is suggested for patients harboring a RAS mutation in comparison with WT patients (median, 26.5 versus 42 months; Fig. 3).

Inclusion of chronic myelomonocytic leukemia into the new category of mixed myelodysplastic syndrome/myeloproliferative neoplasms (12) did not represent a conclusive solution with regard to the diverse clinical phenotypes of this malignancy. Indeed, because the prognostic significance of the distinction between MD-CMML and MP-CMML is still debated upon (5–10) and because the cutoff value of WBC to identify such groups is arbitrary, no general consensus has been reached on this subclassification dilemma. In agreement with previous reports (5, 7–9), our results support its prognostic value because MP-CMML patients had a significantly shorter survival than those with myelodysplastic disease.

We then investigated the biological implication of the myelodysplastic/myeloproliferative distinction, focusing in particular on the role of RAS mutations in chronic myelomonocytic leukemia evolution. To this aim, we conducted a longitudinal monitoring of chronic myelomonocytic leukemia course first by sequencing and then by allele-specific PCR. In fact, among the various assays developed for early detection of specific mutations, tracking of disease progression, and evaluation of molecular response to therapy, which are crucial issues in clinical practice, allele-specific PCR represents a simple, fast, and sensitive method (35, 36). Direct sequencing detected a significantly higher incidence of RAS mutations in MP-CMML patients, consistent with previous reports (1, 5, 7–9, 14). Next, for each mutation identified, we set up a highly selective yet simple and low-cost allele-specific PCR, which, in our experimental conditions, resulted up to 20 times more sensitive than sequencing and did not require expensive equipment and proprietary reagents. Accordingly, at the time of first presentation or referral, allele-specific PCR detected a higher percentage of patients carrying a RAS mutation compared with sequencing. Moreover, allele-specific PCR allowed identification of a mutRAS allele in MD-CMML and MP-CMML patients, whereas sequencing only did in the latter group. Importantly, molecular monitoring done on matched presentation/follow-up pair samples from both MD-CMML patients presenting with a RAS mutation (patients 8 and 33) who experienced progression to myeloproliferative disease, unveiled at the time of myelodysplastic diseaselow levels of the mutations that became predominant in the myeloproliferative phase. Albeit few cases of MD-CMML progressing to myeloproliferative disease have been reported (1, 2, 8, 11), to the best of our knowledge, this is the first study documenting the appearance of RAS mutations in concomitance with chronic myelomonocytic leukemia evolution.

Allele-specific PCR was also able to detect the presence of a mutant allele earlier than sequencing during the course of myeloproliferative disease (patients 24 and 38). Indeed, in the case of patient 24, allele-specific PCR could identify the G12V mutation in the peripheral blood sample obtained during hydroxyurea (HU) treatment 27 months after the first withdrawal that coincided with the initial increase of WBC to more than the 13 × 10⁹/L threshold. Because allele-specific PCR for the G12V allele has shown the lowest sensitivity out of the six allele-specific PCR developed, we cannot rule out the possibility that the mutated clone was already present at very low levels at the time of the first withdrawal. We may also speculate that HU reduced the mutated clone, thus preventing sequencing but not allele-specific PCR (which is far more sensitive) from identifying the G12V allele at the time of the second sample.

Finally, allele-specific PCR provided evidence for the occurrence of the NRAS G12D mutation in a progenitor common to the granulocytic/monocytic and erythroid precursors and in a percentage of cells that reflected the level of mutation detected in total DNA. In these regard, it would be interesting to further explore whether the mutation arises in the stem cells compartment because it was shown very recently in mouse models of myeloproliferative diseases (37, 38). Altogether, our findings suggest the occurrence of a RAS mutated clone in chronic myelomonocytic leukemia that eventually expanded, taking proliferative advantage on the RAS WT cells, and finally led to disease progression along with establishment of a stable myeloproliferative phenotype, which is associated with worse outcome. Notably, in agreement with the findings from the largest chronic myelomonocytic leukemia population ever analyzed (15), also in our cohort of patients the presence of mutRAS significantly correlated with higher WBC counts, whereas association with a shorter survival did not reach statistical significance possibly because of the limited sample size.

From a molecular point of view, albeit the mechanisms at the basis of chronic myelomonocytic leukemia pathogenesis are not fully understood, evidences from in vivo studies along with analysis of patient series suggest that multiple genetic events may contribute to disease development, especially in the proliferative forms. In particular, although RAS mutations seem to be the most commonly identified genetic aberration in MP-CMML, the FLT3-ITD (28) and the JAK2 V617F substitution (29, 30) have also been found in a minority of patients. Indeed, we detected the FLT3-ITD in a case of MD-CMML who rapidly progressed to MP-CMML and then to acute leukemia (patient 31) and the JAK2 V617F substitution in two cases classified as MP-CMML at the time of first presentation (patients 23 and 35; results not shown). Interestingly, the cumulative median survival of all patients (n = 11) presenting with a genetic aberration (mutRAS, FLT3-ITD, or JAK2 V617F, which are mutually exclusive) compared with those without (n = 29) was shorter (25 months versus not reached), and the difference reached statistical significance (P = 0.049). Given that specific inhibitors are currently under investigation for these aberrations (39, 40), their detection is of clinical relevance.

Based on these data, it may be hypothesized that the myelodysplastic/myeloproliferative distinction reflects different molecular subtypes of chronic myelomonocytic leukemia. In particular, as far as the MD-CMML is concerned, it might be supposed that pathogenetic events other than RAS mutations are responsible for its establishment and that acquisition of a secondary event like a RAS mutation or, in a minority of patients, the FLT3-ITD or the JAK2 V617F substitution causes the myeloproliferative phenotype. One possible candidate could be TET2, whose mutations have been recently identified in a wide spectrum of myeloid malignancies (19–26), leading to hypothesize that they are an early event in hematopoietic differentiation and that at least one additional event is required to determine the disease phenotype (19, 27). Moreover, Gelsi-Boyer et al. (14) have recently identified RUNX1 alterations in a series of chronic myelomonocytic leukemia patients with MD-CMML and MP-CMML, whereas RAS mutations were absent in the former group. These results, together with the observation that RAS and RUNX1 aberrations were not mutually exclusive, led the authors to speculate that RUNX1 alterations might be responsible for dysplasia and RAS mutations for myeloproliferation. In contrast, Tyner et al. (41) found RAS mutations in MD-CMML and MP-CMML. One possible explanation for this discrepancy may be the different sensitivity of the methods used. Unfortunately, none of the two groups conducted a longitudinal analysis, neither gave information on the disease course of MD-CMML patients. Clearly, to elucidate the biology of chronic myelomonocytic leukemia, further investigations on larger groups of patients are needed.

In conclusion, albeit not all patients who experienced progression to myeloproliferative phase nor all patients presenting with a de novo myeloproliferative phenotype harbored a mutRAS allele, our results suggest the negative impact of RAS mutations on chronic myelomonocytic leukemia course. Therefore, also in light of the ongoing studies aimed at identifying effective Ras inhibitors (42–45), we believe that detection of RAS mutations in chronic myelomonocytic leukemia could be of clinical relevance. We propose screening by highly selective yet simple and low-cost allele-specific PCR to be implemented as a standard in chronic myelomonocytic leukemia diagnostics and disease monitoring to help clinical decision making.

No potential conflicts of interest were disclosed.

We thank Simona Iacobelli, Ph.D. (Centro Interdipartimentale di Biostatistica e Bioinformatica, Università degli Studi di Roma “Tor Vergata,” Rome, Italy), for the biostatistical advices.

Grant Support: Associazione Italiana contro le Leucemie–Linfomi e Mieloma (sezione di Milano).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.