AUTHORS
Ayalew Tefferi, MD, Matteo Giovanni Della Porta, MD2, Lior Rokach3, Blanca Xicoy, PhD4, Torsten Haferlach, MD, PhD5, Saubia Fathima, MD6, Laura Palomo, PhD7, Ali Alsugair, MBBS8, Giulia Maggioni, MD9, Animesh Pardanani, PhD, MBBS10, Ulrich Germing, MD11, Kaaren Reichard, MD12, Alessia Campagna, MD13, Abhishek Avinash Mangaonkar, MBBS14, Gabriele Todisco, MD15, Rong He, MD16, Anthony Michael Hunter, MD17, Naseema Gangat, MBBS18, Eric Solary, MD19, Cinthya J. Zepeda Mendoza20, Eric Padron, MD21, Mrinal M. Patnaik, MD, MBBS22, Luca Lanino, MD23
AFFILIATIONS
1 Mayo Clinic, Rochester, MN, United States
2 Humanitas University, Milan, Italy
3 Ben-Gurion University of the Negev, Negev, Israel
4 Josep Carreras Leukaemia Research Institute, Barcelona, Spain
5 MLL Munich Leukemia Laboratory, Munich, Germany
6 Mayo Clinic, Rochester, MN, United States
7 Josep Carreras Leukaemia Research Institute, Barcelona, Spain
8 Mayo Clinic, Rochester, MN, United States
9 Humanitas University, Milan, Italy
10 Mayo Clinic, Rochester, MN, United States
11 Heinrich-Heine-University, Dusseldorf, Germany
12 Mayo Clinic, Rochester, MN, United States
13 Humanitas University, Milan, Italy
14 Mayo Clinic, Rochester, MN, United States
15 Humanitas University, Milan, Italy
16 Mayo Clinic, Rochester, MN, United States
17 Emory University, Atlanta, United States
18 Mayo Clinic, Rochester, MN, United States
19 Gustave Roussy Cancer Center, Villejuif, France
20 Mayo Clinic, Rochester, MN, United States
21 Moffitt Cancer Center, Tampa, United States
22 Mayo Clinic, Rochester, MN, United States
23 Humanitas University, Milan, Italy
ABSTRACT
Background: Curative therapy in chronic myelomonocytic leukemia (CMML) is currently limited to allogeneic stem cell transplantation (ASCT). Optimal timing of ASCT is necessary to minimize procedure-related mortality/morbidity and maximize benefit. In this regard, most investigators agree on the early utilization of ASCT in accelerated phase CMML, defined by 10-19% bone marrow (BM) or 5-19% peripheral blood (PB) blasts (CMML-2). The current study is focused on chronic phase CMML (CMML-1; <10% BM and <5% PB blasts) and was designed to develop a molecularly simplified risk model that incorporates the recently introduced BLAST clinical score (Blood 2025;146:874) and the mutation-independent cytogenetic risk classifier (Cyto-CMML; EHA 2026).
Methods: A Mayo Clinic discovery cohort and a multicenter validation cohort were considered. Diagnoses were according to ICC (Blood 2022;140:1200); cases with NPM1MUT were excluded. NGS platforms were used to curate mutation information. Overall survival (OS) and cumulative incidence of blast transformation (BT) calculations were adjusted for ASCT and competing risk, respectively. A machine-learning approach was undertaken to select prognostically informative mutation combinations.
Results: The Mayo Clinic discovery cohort included 592 patients with CMML-1 (median age 72 years; males 70%). BLAST clinical risk categories were low (38%), intermediate (36%), and high (26%). Genetic characteristics included adverse karyotype (AK) per Cyto-CMML (4%), TET2MUT (57%), SRSF2MUT (47%), ASXL1MUT (45%), RUNX1MUT (17%), NRASMUT (14%), KRASMUT (11%), CBLMUT (10%), PHF6AMUT (8%), SETBP1MUT (7%), U2AF1MUT (7%), JAK2MUT (7%), DNMT3AMUT (7%), ZRSR2 (5.4%), EZH2 (4.7%), IDH2 (4.4%), SF3B1 (4.4%), and PTPN11 (2.7%); other mutations were less frequent and included IDH1 (1.9%), TP53 (1.7%), BCOR (1.4%), FLT3 (1.2%), and CEBPA (1.2%). Median f/u was 5.9 years and included 97 BTs, 92 ASCTs and 405 deaths.
Multivariable analysis (MVA) that included mutations with frequencies of ≥2% identified BLAST high-risk (HR 3.6; p<0.01), BLAST intermediate-risk (HR 1.9; p<0.01), AK (HR 5.2; p<0.01), ASXL1MUT (HR 1.6; p<0.01), TET2MUT (HR 0.6; p<0.01), and U2AF1MUT (HR 1.5; p=0.03) as independent predictors of OS; introduction of less frequent (<2%) mutations in the MVA model revealed additional significance for IDH1 (p=0.01), BCOR (p=0.04), CEBPA (p=0.01), FLT3 (p=0.02), but not for TP53 (p=0.2). A machine-learning approach based on decision forest revealed significant modulation of the prognostic impact from ASXL1MUT and TET2MUT by their co-mutation pattern: TET2MUT/ASXL1WT favorable (reference); ASXL1MUT/TET2WT unfavorable (HR 3.2; p<0.01); other ASXL1/TET2 permutations intermediate (HR 1.9; p<0.01).
The above-underlined ASXL1/TET2 co-mutation pattern was subsequently included in a MVA that identified AK (HR 5.1; p<0.01; 4 points), BLAST high-risk (HR 3.5; p<0.01; 3 points), BLAST intermediate-risk (HR 1.9; p<0.01; 1 point), ASXL1MUT/TET2WT (HR 2.6; p<0.01; 2 points), and TET2MUT/ASXL1MUT or TET2WT/ASXL1WT (HR 1.7; p<0.01; 1 point) as independent predictors of OS. This information was used to develop a molecularly-simplified (only 2 mutations considered) “BLAST-CMML-1” risk model: high (≥5 points; N=76; median 1 year), intermediate-2 (2-4 points; N=262; median 2.7 years), intermediate-1 (1 point; N= 138; median 4.8 years), and low (0 points; N=116; median 6.8 years) risk categories (inter-group p<0.02). Predictive performance of BLAST-CMML-1 (5-year AUC 0.79; C-index 0.70) compared favorably with CPSS-mol (AUC 0.69; C-index 0.64), BLAST clinical (AUC 0.74; C-index 0.62), and BLAST-mol (AUC 0.76; C-index 0.67) [p<0.01 for AUC and C-index comparisons of BLAST-CMML-1 vs. all other models]. Five-year BT risk was 18% and independently predicted by FLT3MUT (HR 4.6; p<0.01), ASXL1MUT (HR 1.5; p=0.04), and JAK2MUT (HR 2.4; p<0.01). The OS advantage of ASCT was most apprent for BLAST-CMML-1 high and intermediate-2 risk categories. The new model was externally validated by an international cohort of 969 patients with corresponding median OS of 1.9, 2.8, 4.3, and 8.4 years (5-years AUC 0.74).
Conclusion: BLAST-CMML-1 stands out from previous molecular risk models by specifically targeting chronic phase disease (CMML-1) and relying on only two mutations (TET2 and ASXL1) that are frequently occuring. The model is simple, validated, and comparable to other more complex molecular risk models in its predictive performance.
DISCLOSURES
Ayalew Tefferi, MD: Disc Medicine - Research Funding