Key Points
-
•
IPSS-M stratifies patients with MD-CMML/MP-CMML into well-defined risk categories with distinct outcomes in terms of OS and AML progression.
-
•
IPSS-M shows comparable prognostic performance to iCPSS and improved discrimination compared with CPSS-Mol.
Visual Abstract
Abstract
Prognostication in chronic myelomonocytic leukemia (CMML) remains a challenge because of the biological complexity and variable clinical course of the disease. This study aimed to evaluate the prognostic utility of the Molecular International Prognostic Scoring System (IPSS-M) in CMML and its applicability across the myelodysplastic (MD) and myeloproliferative (MP) subsets of the disease. We conducted a multicenter, retrospective study including 511 patients diagnosed with CMML. Clinical, cytogenetic, and molecular data were collected at diagnosis, including targeted next-generation sequencing. Patients were stratified using IPSS-M, the molecular CMML-specific prognostic scoring system (CPSS-Mol), and the recently developed international CPSS (iCPSS). IPSS-M effectively stratified patients into risk groups with significantly different overall survival (OS) and cumulative incidence of acute myeloid leukemia (AML) progression. Discrimination was maintained after merging overlapping intermediate-risk categories, yielding a 4-tier model with a C-index of 0.678 for OS and 0.628 for AML progression, respectively. This model retained its prognostic performance in both MD-CMML and MP-CMML subsets, with a higher discrimination for OS in the MD-CMML group. When compared with CPSS-Mol and iCPSS, adapted IPSS-M showed comparable prognostic performance to iCPSS and improved discrimination compared with CPSS-Mol. These findings support the applicability of an adapted IPSS-M to CMML, extending its use beyond myelodysplastic syndromes and highlighting its potential utility in guiding clinical decision-making and therapeutic strategies. Moreover, this study provides an external validation of the iCPSS in an independent and genetically well-characterized CMML cohort, reinforcing its clinical utility.
Introduction
Chronic myelomonocytic leukemia (CMML) is a clonal hematologic neoplasm characterized by overlapping features of myelodysplastic syndromes (MDS) and myeloproliferative neoplasms (MPN).1 CMML is a heterogeneous disease, traditionally classified into 2 variants based on white blood cell (WBC) count: MD-CMML (WBC, <13 × 109/L) and MP-CMML (WBC, ≥13 × 109/L).2 Prognostication in CMML remains a challenge because of its biological complexity and variable clinical course, requiring robust risk stratification tools to guide treatment decisions.
Several scoring systems that are specific to CMML have been proposed over the years. The molecular CMML-specific prognostic scoring system (CPSS-Mol) refines previous models by integrating molecular data to classify patients into 4 risk groups.3 This index incorporates key clinical and genetic risk factors, including the percentage of bone marrow (BM) blasts, WBC count, transfusion dependency, cytogenetic abnormalities, and recurrent somatic mutations in ASXL1, NRAS, RUNX1, and SETBP1. The CPSS-Mol has demonstrated utility in predicting overall survival (OS) and leukemic transformation, making it a valuable tool for risk-adapted therapeutic strategies.
More recently, the international CPSS (iCPSS) has been developed in the largest CMML cohort reported so far, including 3182 patients.4 This prognostic model was generated using machine learning algorithms in a large international training cohort and subsequently validated in an independent prospective cohort with centralized molecular analysis. To build this scoring system, a broader spectrum of gene mutations was assessed compared with the previous index. The iCPSS stratifies patients into 5 risk categories by integrating hematologic parameters (WBC, hemoglobin levels [Hb], platelet counts [PLT], and BM blasts), cytogenetic abnormalities, and mutations in 9 genes (ASXL1, DNMT3A, EZH2, RUNX1, SETBP1, STAG2, TET2, TP53, and U2AF1). Given its very recent development, independent external validation studies in additional CMML cohorts are still lacking.4 For patients with MDS, the Molecular International Prognostic Scoring System (IPSS-M) provides a refined classification into 6 risk categories by incorporating conventional variables including BM blast percentage, Hb, PLT, and cytogenetic abnormalities, alongside a comprehensive mutational profile encompassing 31 genes.5 Mutations across these genes are common in other myeloid neoplasms besides MDS, including CMML.6,7 In fact, the IPSS-M was derived from a large cohort of 2957 patients, predominantly MDS patients, but also included 272 patients with MD-CMML, suggesting its potential applicability in this subset of patients. However, its performance in the broader CMML population, including MP-CMML, remains uncertain.
Given the growing reliance on molecular profiling for risk assessment, it is critical to evaluate the utility of the IPSS-M in CMML and compare it with established CMML-specific models such as CPSS-Mol and iCPSS. This study aims to assess the prognostic value of IPSS-M in CMML and explore its applicability across both the MD and MP variants of the disease.
Methods
Patients
Patients diagnosed with CMML according to the 2022 World Health Organization (WHO) classification, with a minimum follow-up of 3 months, were included in this multicenter retrospective study.1 Mutational analysis performed at diagnosis was required for inclusion. This study was approved by the Hospital Germans Trias i Pujol Ethics Committee (reference number CEI: PI-24-246). Data were collected with patient consent, under protocols approved by the institutional research boards of the participating institutions, in accordance with the Declaration of Helsinki.
Clinical and genetic data
Clinical and laboratory data were collected at the time of diagnosis, including sex, age, Hb, total WBC counts, neutrophil and monocyte counts, PLT, percentage of BM blasts (including promonocytes), and transfusion dependency. Genetic analysis was conducted at each participating center as part of the diagnostic process. Collected data included karyotypes, obtained by conventional G-banding analysis, and mutational profiles assessed by targeted next-generation sequencing (NGS). A set of 29 genes was commonly analyzed across all institutions (ASXL1, BRAF, CALR, CBL, CEBPA, CSF3R, DNMT3A, ETV6, EZH2, FLT3, GATA2, IDH1, IDH2, JAK2, KIT, KRAS, MPL, NPM1, NRAS, PTPN11, RUNX1, SETBP1, SF3B1, SRSF2, TET2, TP53, U2AF1, WT1, and ZRSR2). Supplemental Table 1 summarizes the genes analyzed across participating centers and details the IPSS-M molecular variables missing in each institution, including the corresponding subsets of patients affected by missing variables. Clinical outcomes included the date of death from any cause or last follow-up, the date of hematopoietic stem cell transplant (HSCT), and the date of acute myeloid leukemia (AML) transformation.
Prognostic risk stratification
Patients were classified into different risk categories according to IPSS-M, CPSS-Mol, and iCPSS.3, 4, 5Mutations in genes included in CPSS-Mol and iCPSS were available in all patients. Mutations in some of the genes integrated in the IPSS-M were missing for some patients (supplementary Table 1), but because this model accounts for missing variables, all patients could be assigned to a risk group.
Statistical analysis
A descriptive analysis was performed to summarize the main patient characteristics. OS was calculated using the Kaplan-Meier method, and for comparisons between groups, the log-rank test was used. The cumulative incidence of progression to AML was analyzed using competing risks (considering nonprogression mortality as a competing event) and compared by the Gray test. Hazard ratios were estimated by the Cox model (for OS) and by the Fine and Gray model (for cumulative incidence of progression to AML). The discriminative ability of the CPSS-Mol, IPSS-M, and iCPSS models was evaluated using the concordance index (Harrell C-index). Two-sided values of P < .05 were considered statistically significant. All statistical analyses were performed using R version 4.4.1, and the Sankey diagram was made with SankeyMATIC.
Results
Patient characteristics
A total of 511 patients diagnosed with CMML from 19 different institutions of the Spanish Group of Myelodysplastic Syndromes (GESMD) and from the Munich Leukemia Laboratory were included in the study. The median follow-up for alive patients was 2.18 years (range, 0.25-15.35). The characteristics of the patients are summarized in Table 1. The median age at diagnosis was 73 years (range, 27-95), with a male-to-female ratio of 2.3:1. According to the 2022 WHO classification, 410 (80.2%) patients were classified as CMML-1 and 95 (18.6%) as CMML-2. Additionally, we included 6 (1.2%) patients with NPM1 mutation who had been diagnosed with CMML-2 before 2022 but who met the diagnostic criteria for AML with NPM1 mutation according to the current 2022 WHO classification. As per the French-American-British (FAB) classification, 342 (66.9%) patients were classified as having MD-CMML and 169 (33.1%) patients as having MP-CMML. The median OS of the cohort was 4.51 years (95% confidence interval [CI], 3.89-5.61), with a 5-year OS of 49% (95% CI, 43-54). Regarding AML progression, 97 (19%) patients progressed to AML, with an incidence of 22% (95% CI, 18-26) at 5 years. Of note, 38 patients (7%) fulfilled current WHO 2022 criteria for oligomonocytic CMML (monocyte count between 0.5 and 1.0 × 10⁹/L). Comparison with the remaining CMML cohort showed expected differences in WBC and monocyte counts, but no significant differences in OS, AML progression, or distribution of IPSS-M and iCPSS risk categories (supplemental Table 2).
Table 1.
Main characteristics of the cohort of patients (N = 511)
| Variable | Median (range)/n (%) |
|---|---|
| Age, median (range), y | 73 (27-95) |
| Sex, n (%) | |
| Male | 358 (70.1) |
| Female | 153 (29.9) |
| FAB, n (%) | |
| MD | 342 (66.9) |
| MP | 169 (33.1) |
| WHO 2022, n (%) | |
| CMML-1 | 410 (80.2) |
| CMML-2 | 101 (19.8) |
| Hb, median (range), g/dL | 11.80 (4.90-17.80) |
| WBC, median (range), × 109/L | 9 (2.40-93.67) |
| Neutrophils, median (range), × 109/L | 4.06 (0.10-54.73) |
| Monocytes, median (range), × 109/L | 1.90 (0.52-36.60) |
| Monocytes, median (range), % | 22 (10-71) |
| Platelets, median (range), × 109/L | 126 (7-1952) |
| Blasts + promonocytes in BM, median (range), % | 4.5 (0-19) |
| Transfusion dependency, n (%) | |
| Yes | 98 (19.2) |
| No | 413 (80.0) |
| Allogeneic stem cell transplant, n (%) | |
| Yes | 50 (9.8) |
| No | 461 (90.2) |
| AML progression, n (%) | 97 (19.0) |
| Median, y | Not reached |
| Cumulative incidence at 5 y (95% CI), % | 22 (18-26) |
| OS median (95% CI), y | 4.51 (3.89-5.61) |
| At 5 years (95% CI) | 49 (43-54) |
Cytogenetic analysis showed that 80 (15.7%) patients presented with an altered karyotype. According to the CMML-specific cytogenetic risk classification,8 446 (87.2%) patients were classified as low risk (431 patients with a normal karyotype and 15 with isolated –Y), 30 (5.9%) as intermediate risk, and 36 (6.0%) as high risk (17 patients with +8, 13 with -7/del(7q), and 5 with complex karyotypes). Mutational analysis via targeted NGS revealed that most patients (n = 487 [95.3%]) had ≥1 mutation. The most frequently mutated genes (>5%) were TET2 (64.0%), ASXL1 (36.6%), SRSF2 (36.6%), RUNX1 (17.4%), CBL (12.5%), NRAS (11.7%), KRAS (9.8%), ZRSR2 (9.8%), EZH2 (8.4%), SETBP1 (7.6%), JAK2 (6.5%), DNMT3A (6.1%), PHF6 (5.1%), and NF1 (5.1%; supplemental Figures 1 and 2). Of the 327 patients with TET2 mutations, 192 (58.7%) had multiple TET2 mutations, and 141 (43.1%) had an SRSF2 mutation; both features have been associated with CMML. Of note, 10 (2.0%) patients showed TP53 mutations (all of them single mutations), and only 1 of these had a variant allele frequency >50% suggestive of a biallelic inactivation; this patient had an altered karyotype with a del(4q) and a del(5q).
Applicability of IPSS-M in CMML
Patients were stratified according to the IPSS-M and categorized into 6 risk groups: very low (VL; n = 110 [21.5%]), low (L; n = 144 [28.2%]), moderate low (ML; n = 73 [14.3%]), moderate high (MH; n = 57 [11.2%]), high (H; n = 80 [15.7%]), and very high (VH; n = 47 [9.2%]). These groups showed significantly different OS and incidence of AML transformation (P < .001; supplemental Figure 3). Patients from the L and ML risk groups showed overlapping curves, suggesting suboptimal calibration of the original IPSS-M categories in CMML; therefore, these groups were combined into a single L/ML group (n = 217 [39.4%]). Similarly, the MH and H risk groups were merged into a single MH/H group (n = 137 [24.9%]), resulting in an adapted 4-tier IPSS-M model for CMML. Following this new stratification, the OS and cumulative incidence of AML progression at 3 years were as follows: VL (OS, 90% [81%-95%]; AML incidence, 2% [0%-8%]), L/ML (OS, 69% [61%-76%]; AML incidence, 16% [11%-22%]), MH/H (OS, 43% [33%-52%]; AML incidence, 24% [16%-32%]), and VH (OS, 24% [12%-38%]; AML incidence, 42% [27%-56%]; Figure 1). A sensitivity analysis censoring patients at the time of allogeneic HSCT yielded comparable survival results (data not shown). Overall, IPSS-M demonstrated high discriminative capacity for OS and AML progression in patients with CMML. Using the restratified IPSS-M (4 groups), the concordance index (C-index) for OS was 0.678 (95% CI, 0.641-0.715), and for the risk of AML progression, the C-index was 0.628 (95% CI, 0.541-0.678).
Figure 1.
Risk stratification of patients with CMML according to IPSS-M reclassified into 4 risk groups. (A) OS curves; (B) AML incidence curves.
Applicability of IPSS-M in MD-CMML and MP-CMML
Next, we assessed the applicability of IPSS-M in MD-CMML and MP-CMML subtypes using the restratified IPSS-M (4 groups).
In MD-CMML (n = 342), patients were classified into VL (n = 85 [24.8%]), L/ML (n = 171 [50.0%]), MH/H (n = 67 [19.6%]), and VH (n = 19 [5.6%]) risk groups, with significant differences in both OS (P < .001) and risk of AML transformation (P < .001; Figure 2). The 3-year OS and cumulative incidence of AML progression for MD-CMML were as follows: VL (OS, 92% [81%-97%]; AML incidence, 3% [1%-10%]), L/ML (OS, 73% [64%-80%]; AML incidence, 13% [8%-19%]), MH/H (OS, 44% [29%-57%]; AML incidence, 24% [14%-37%]), and VH (OS, 28% [10%-50%]; AML incidence, 43% [20%-64%]). The discriminative power of reclassified IPSS-M for MD-CMML was high, with a C-index for OS of 0.689 (95% CI, 0.642-0.736), slightly superior to that in the complete CMML cohort, and a C-index for the risk of AML progression of 0.627 (95% CI, 0.519-0.695).
Figure 2.
Risk stratification of patients with MD-CMML according to IPSS-M, reclassified into 4 risk groups. (A) OS curves; (B) AML incidence curves.
Patients with MP-CMML (n = 169) were classified into VL (n = 25 [15.0%]), L/ML (n = 46 [27.2%]), MH/H (n = 70 [41.4%]), and VH (n = 28 [16.6%]) risk groups. These also showed significant differences in both OS (P < .001) and risk of AML transformation (P = .013; Figure 3). The 3-year OS and cumulative incidence of AML progression for MP-CMML were as follows: VL (OS, 82% [58%-93%], AML incidence, 0%), L/ML (OS, 55% [37%-69%], AML incidence, 27% [14%-42%]), MH/H (OS, 42% [29%-55%]; AML incidence, 23% [13%-34%]); and VH (OS, 21% [7%-40%]; AML incidence, 41% [21%-59%]). IPSS-M was able to discriminate OS in patients with MP-CMML, with a C-index for OS of 0.616 (95% CI, 0.551-0.681). However, for AML progression, the C-index could not be confidently assessed because of the low number of events in the L risk groups.
Figure 3.
Risk stratification of patients with MP-CMML according to IPSS-M, reclassified into 4 risk groups. (A) OS curves; (B) AML incidence curves.
Comparison of IPSS-M with other CMML molecular scoring systems
After confirming the applicability of the adapted 4-tier IPSS-M in CMML, we compared its performance with other CMML-specific scoring systems.
First, patients were stratified using the CPSS-Mol schema into 4 risk groups: low (n = 107 [20.9%]), intermediate-1 (n = 133 [26.0%]), intermediate-2 (n = 158 [30.9%]), and high (n = 113 [22.1%]), which showed significantly different OS and risk of AML transformation (P < .001; Figure 4A-B). Of note, the OS and AML curves for the low and intermediate-1 groups overlapped, showing no difference between the outcomes of these 2 groups. The 3-year OS and cumulative incidence of AML evolution were as follows: low (OS, 78% [67%-86%]; AML incidence, 4% 1%-10%]), intermediate-1 (OS, 77% [67%-84%]; AML incidence, 8% [4%-15%]), intermediate-2 (OS, 55% [46%-64%]; AML incidence, 21% [15%-28%]), and high (OS, 38% [28%-49%]; AML incidence, 35% [26%-45%]).
Figure 4.
Risk stratification of patients with CMML according to CPSS-Mol and iCPSS. (A) CPSS-Mol OS curves; (B) CPSS-Mol AML incidence curves; (C) iCPSS OS curves; (D) iCPSS AML incidence curves. Int, intermediate.
Second, patients were stratified according to the recently developed iCPSS, which classified patients into 5 risk groups: VL (n = 135 [26.4%]), low (n = 212 [41.5%]), intermediate (n = 81 [15.9%]), high (n = 56 [11.0%]), and VH (n = 27 [5.3%]). iCPSS successfully stratified patients into 5 groups with significantly different OS (P < .001; Figure 4C). Similarly, iCPSS identified distinct groups of patients with different risks of AML progression (Figure 4D).
Figure 5 shows how patients classified with CPSS-Mol and iCPSS were restratified into different risk groups according to reclassified IPSS-M (4 groups; Figure 5; supplemental Figure 4). Finally, we compared the performances of the 3 scoring systems across the entire cohort using the Harrell C-index. (Figure 6; supplemental Table 3). All scoring systems performed better in OS discrimination than in the risk of AML evolution. For OS, IPSS-M (4 groups) and iCPSS showed similar discriminative capacity and equally showed improved OS discrimination over the CPSS-Mol, as demonstrated by a 4.9% (0.678 vs 0.629) and 3.3% (0.662 vs 0.629) increase in C-index, respectively. Regarding cumulative incidence of AML progression, all 3 models demonstrated a comparable ability to predict leukemic transformation, with similar C-index values across the cohort (0.628 for IPSS-M, 0.633 for CPSS-Mol, and 0.625 for iCPSS; Figure 6).
Figure 5.
Sankey diagram showing the reclassification of patients with CMML according to the different risk scores. Int, intermediate.
Figure 6.
Bar plot showing the C-index values of reclassified IPSS-M, CPSS-Mol, and iCPSS across both end points. (A) OS; (B) risk of AML progression.
Discussion
Predicting outcomes in CMML remains challenging because of its biological complexity and heterogeneous clinical behavior. The integration of molecular data into prognostic models is essential because somatic mutations significantly influence disease phenotype, progression risk, and response to treatment.3,6,9,10 Accurate risk stratification tools are crucial to guide therapeutic decisions, particularly in an era of increasingly personalized medicine.
In our multicenter cohort of 511 patients with CMML, we evaluated the performance of the IPSS-M,5 a model originally developed for MDS, and assessed its applicability in CMML as a whole and in its 2 main clinical subtypes: MD-CMML and MP-CMML. When applied to the full CMML cohort, the IPSS-M effectively stratified patients into 6 distinct risk groups with significantly different OS and risk of AML transformation. Approximately half of the patients fell into the 2 lowest-risk categories, whereas only one-quarter were classified within the 2 highest-risk groups, indicating that the model captures the wide biological spectrum of the disease.
To improve interpretability and address overlapping survival curves, particularly between the L and ML, and MH and H risk groups, we reclassified patients into 4 risk categories. Importantly, the need to collapse adjacent IPSS-M categories suggests that the original 6-tier IPSS-M is not optimally calibrated for CMML biology. Therefore, rather than representing a direct application of the original MDS-derived IPSS-M, our findings support the use of an adapted 4-tier IPSS-M model in CMML. This restratified IPSS-M retained good discriminative power for both OS (C-index, 0.678) and AML progression (C-index, 0.628), supporting its applicability in CMML.
When the reclassified IPSS-M was applied specifically to patients with MD-CMML, three-quarters of patients were allocated to the lower-risk groups (VL and L/ML), aligning well with the known more indolent clinical course of this subtype. Importantly, although the IPSS-M does not include WBC, on which the FAB classification is based, it still accurately reflects the more favorable prognosis generally associated with MD-CMML. The discriminative ability of the model in this subgroup was slightly superior to that observed in the overall cohort (C-index for OS, 0.689), which is consistent with the IPSS-M’s original development in patients with MDS. However, its performance was still lower than reported by Bernard et al5 in the original MDS cohort.
In patients with MP-CMML, more than half were classified into the higher risk categories (MH/H and VH), reflecting the more aggressive nature of this subtype. The IPSS-M was able to significantly discriminate OS across risk groups (C-index for OS, 0.616), although its performance for predicting AML progression was limited by the lower number of transformation events in the lower-risk categories, limiting the robustness of AML progression analyses in this subgroup and warranting cautious interpretation.
We also compared the IPSS-M with 2 CMML-specific prognostic models: the CPSS-Mol and the recently developed iCPSS.3,4 The CPSS-Mol is a well-established and widely used prognostic model in CMML; however, one of its main limitations is that it includes mutations in only 4 genes, which may limit its ability to capture the full genetic complexity of the disease. In contrast, the iCPSS is a novel prognostic tool that integrates clinical, hematologic, and molecular variables and was developed using machine learning algorithms on a large, multicenter CMML data set. Its design aims to overcome limitations of previous scores by incorporating more refined and comprehensive predictors of outcome. Importantly, iCPSS represents a CMML-specific prognostic model developed for this disease and has demonstrated superior prognostic discrimination compared with previous scoring systems, including IPSS-M, in large international cohorts.4 In our cohort, iCPSS successfully stratified patients into 5 groups with significantly different OS and risk of AML progression, showing a predictive performance comparable to the adapted IPSS-M. Of note, this study provides an external validation of the iCPSS in an independent and genetically well-characterized CMML cohort, reinforcing its clinical utility. As iCPSS has only recently been developed, independent validation data are still limited, making external assessments in additional CMML cohorts particularly relevant.
All 3 models effectively stratified patients by OS. However, both IPSS-M and iCPSS showed superior discrimination for OS compared with the CPSS-Mol, with an increase of 4.9% and 3.3% in C-index, respectively. Regarding prediction of AML progression, the 3 prognostic models demonstrated overall comparable performance, with similar discriminative capacity across the cohort. Overall, these findings support the utility of both adapted IPSS-M and iCPSS as robust prognostic tools for stratifying clinical outcomes in CMML, including both OS and risk of AML progression. Of note, although a small proportion of patients fulfilled criteria for oligomonocytic CMML according to current classifications, their inclusion did not appear to affect the prognostic performance of IPSS-M or iCPSS in our cohort.
Our study has certain limitations. First, its retrospective design may introduce selection biases. In addition, the multicenter nature of the cohort, while increasing sample size, generalizability, and reflecting real-world clinical practice, also introduces heterogeneity in diagnostic approaches, molecular testing platforms, cytogenetic assessment procedures, and treatment strategies across participating institutions. Such variability may have influenced the estimation of prognostic performance and should be considered when interpreting relatively small differences in C-index values between prognostic models. Additionally, the absence of centralized molecular testing and potential heterogeneity in the genetic panels used could have affected the uniformity of IPSS-M scoring, particularly regarding the assessment of residual IPSS-M genes and TP53 allelic status in a subset of patients. However, the main IPSS-M genes were analyzed in all patients (except for the rare KMT2A-PTD event, in a proportion of them). In addition, our study represents a retrospective local implementation in a smaller cohort and does not reproduce the centralized molecular profiling or prospective validation strategy of the original iCPSS study. Therefore, comparisons between prognostic models should be interpreted cautiously. Finally, treatment-related variables, including exposure to hypomethylating agents and other disease-directed therapies, were not uniformly available and therefore could not be incorporated into the analyses. Although HSCT data were available, survival analyses were not censored at transplant because this approach was considered more representative of overall patient outcomes in this retrospective real-world cohort.
An additional consideration relates to the inclusion of 6 NPM1-mutated patients that would currently be classified as AML according to WHO 2022 and International Consensus Classification (ICC) criteria. These patients fulfilled CMML diagnostic criteria at the time of diagnosis and were retained in the analyses to preserve consistency with the original design and intended application of the evaluated prognostic models, including IPSS-M, which incorporates NPM1 as an adverse molecular variable. Interestingly, NPM1-mutated patients were assigned to high-risk categories by both IPSS-M and iCPSS, and 4 of the 6 patients subsequently progressed to AML, supporting the ability of molecularly informed models to identify biologically aggressive disease.
Overall, in this large multicenter cohort of patients with CMML, the adapted IPSS-M provided clinically meaningful risk stratification for both OS and AML progression. These findings support its role as a practical and informative complementary prognostic tool in CMML while also providing a harmonized risk stratification framework applicable across myeloid neoplasms.
Conflict-of-interest disclosure: The authors declare no competing financial interests.
Acknowledgments
Authorship
Contribution: L.P., L.Z., and B.X. designed the study; M. Morgades provided statistical analyses; L.P. and M. Morgades performed data analysis; L.P., M. Meggendorfer, M.D.-B., H.P., J.F., G.R., M.A.I., M.T., F.L.-C., A.A., J.C.C., T.B., J.V., J.R., V.R., P.G., E.C., C.N., A.G.-F., T.H., S.C.-D., E.A., M.P., R.C., M. Calabuig, P.A., M.R., G.A., C.L., M. Cabezón, N.R.-X., L.Z., B.X. and The Spanish Group of MDS collected clinical data; L.P., M.M., L.Z., and B.X. drafted the manuscript; and all authors revised and approved the manuscript.
Footnotes
Original data are available from the authors, Laura Palomo (lpalomo@iconcologia.net) and Lurdes Zamora (lzamora@iconcologia.net), on request.
The full-text version of this article contains a data supplement.
Supplementary Material
References
- 1.Khoury JD, Solary E, Abla O, et al. The 5th edition of the World Health Organization classification of haematolymphoid tumours: myeloid and histiocytic/dendritic neoplasms. Leukemia. 2022;36(7):1703–1719. doi: 10.1038/s41375-022-01613-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Bennett JM, Catovsky D, Daniel MT, et al. Proposals by the French-American-British Cooperative Leukaemia Group. Br J Haematol. 1994;87(4):746–754. doi: 10.1111/j.1365-2141.1994.tb06734.x. [DOI] [PubMed] [Google Scholar]
- 3.Elena C, Gallì A, Such E, et al. Integrating clinical features and genetic lesions in the risk assessment of patients with chronic myelomonocytic leukemia. Blood. 2016;128(10):1408–1417. doi: 10.1182/blood-2016-05-714030. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lanino L, Hunter AM, Gagelmann N, et al. iCPSS Alliance A molecular-based ecosystem to improve personalized medicine in chronic myelomonocytic leukemia. J Clin Oncol. 2026;44(17):1608–1623. doi: 10.1200/JCO-25-02116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Bernard E, Tuechler H, Greenberg PL, et al. Molecular international prognostic scoring system for myelodysplastic syndromes. NEJM Evid. 2022;1(7) doi: 10.1056/EVIDoa2200008. [DOI] [PubMed] [Google Scholar]
- 6.Itzykson R, Kosmider O, Renneville A, et al. Prognostic score including gene mutations in chronic myelomonocytic leukemia. J Clin Oncol. 2013;31(19):2428–2436. doi: 10.1200/JCO.2012.47.3314. [DOI] [PubMed] [Google Scholar]
- 7.Palomo L, Meggendorfer M, Hutter S, et al. Molecular landscape and clonal architecture of adult myelodysplastic/myeloproliferative neoplasms. Blood. 2020;136(16):1851–1862. doi: 10.1182/blood.2019004229. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Such E, Cervera J, Costa D, et al. Cytogenetic risk stratification in chronic myelomonocytic leukemia. Haematologica. 2011;96(3):375–383. doi: 10.3324/haematol.2010.030957. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Duchmann M, Yalniz FF, Sanna A, et al. Prognostic role of gene mutations in chronic myelomonocytic leukemia patients treated with hypomethylating agents. EBioMedicine. 2018;31:174–181. doi: 10.1016/j.ebiom.2018.04.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Coston T, Pophali P, Vallapureddy R, et al. Suboptimal response rates to hypomethylating agent therapy in chronic myelomonocytic leukemia; a single institutional study of 121 patients. Am J Hematol. 2019;94(7):767–779. doi: 10.1002/ajh.25488. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.