Revised Mayo Molecular Risk Model for Chronic Myelomonocytic Leukemia (MMM version 2.0): A Combined Clinical, Cytogenetic, and Mutational Platform

Background: Previously published risk variables for chronic myelomonocytic leukemia (CMML) included age, gender, CMML subtype, red blood cell (RBC) transfusion status, hemoglobin, leukocyte, monocyte, lymphocyte and platelet counts, immature myeloid cells, peripheral blood (PB) or bone marrow blast percentage, and mutations including ASXL1, RUNX1, NRAS, and SETBP1. Recent studies have reported on additional prognostically favorable (TET2) or unfavorable (DNMT3A, ASXL1, NRAS, SETBP1) mutations. In the current study, we took advantage of a clinically well-characterized and molecularly annotated cohort of CMML patients, in order to revisit the prognostic interaction between the aforementioned risk factors, with emphasis on mutations.

Methods: The current study was conducted under an institutional review board approved protocol that allowed retrospective collection and analysis of data from Mayo Clinic patient records. Mutations were screened by multi-gene next-generation sequencing (NGS). Diagnostic criteria were according to the International Consensus Classification (Blood 2022; 140:1200). Conventional statistical methods were employed (JMP Pro 17.0.0 software SAS Institute, Cary, NC, USA). Survival analysis was censored for allogeneic stem cell transplantation (ASCT). Receiver Operating Characteristic (ROC) plots and Akaike Information Criterion (AIC) were used to compare predictive performance of the new model (MMMv2) vs. the Mayo Molecular Model (MMM) vs. the CMML-specific prognostic scoring system (CPSS).

Results: The study cohort included 416 patients with CMML that were fully annotated for mutations. Cytogenetic information was available in 404 patients. At a median follow-up of 18 months, 302 (72%) deaths, 72 (18%) leukemic transformations, and 44 (11%) ASCTs were documented. Age-adjusted multivariable analysis (MVA) of mutations identified the following as risk factors: DNMT3A (HR 5.5; p<0.01), PHF6 (HR 0.3; p<0.01), TET2 (HR 0.6; p<0.01), PTPN11 (HR 2.5; p=0.01), NRAS (HR 1.6; p<0.01), ASXL1 (HR 1.3; p=0.02), and SETBP1 (HR 1.8; p=0.01). A separate MVA of clinical and cytogenetic variables identified RBC transfusion need, leukocyte count ≥13 x 10⁹/L, peripheral blood blast ≥1%, abnormal karyotype, older age, and male sex to be independently associated with inferior survival (p<0.04 in all instances).

All-inclusive MVA identified the following as independent risk factors for survival (p <0.01): DNMT3A mutation (HR 3.7; 95% CI 2.0-6.7), absence of PHF6 mutation (HR 2.8; 1.3-5.8), RBC transfusion need (HR 2.3; 1.7-3.1), leukocyte count ≥13 x 10⁹/L (HR 2.0; 1.5-2.6), abnormal karyotype (HR 1.6; 1.2-2.2), and absence of TET2 mutation (HR 1.4,1.1-1.8). An HR based prognostic model was subsequently constructed by assigning 3 points to DNMT3A mutation, 2 points each for RBC transfusion need, leukocyte count ≥13 x 10⁹/L, and absence of PHF6 mutation and one point each for abnormal karyotype and absence of TET2 mutation, resulting in a 4-tiered survival risk model (MMMv2): low risk (0-2 points; N=103) with median survival of 74 months; intermediate-1 (3 points; N=65) with median 37 months; intermediate-2 (4-5 points; N=126) with median 19 months; and high (6 points; N=88) with median 12 months (p<0.01).

Independent genetic risk factors associated with inferior leukemia-free survival included DNMT3A (HR 5.5), PHF6 (HR 0.2), abnormal karyotype (HR 2.0), PTPN11 (HR 5.2), ASXL1 (HR 2.0), and BCOR (HR 8.7). Leukocyte count ≥13 x 10⁹/L (p=0.01) but not RBC transfusion need (p=0.2) was additionally significant. However, presence of immature myeloid cells (p<0.01) and circulating blasts ≥1% (p<0.01) were prognostically more relevant than leukocyte count ≥13 x 10⁹/L and independently predicted leukemic transformation, along with DNMT3A, PTPN11 and ASXL1 mutations. Predictive performance was superior for MMMv2 (AUC 87, AIC 137) vs. MMM (AUC 74, AIC 194) vs. CPSS (AUC 76, AIC 177).

Conclusion: The revised Mayo Molecular Model (MMMv2) provides a contemporary risk model for CMML that recognizes DNMT3A as the most unfavorable and PHF6 as the most favorable mutation and confirms the mutation-independent prognostic contribution from red cell transfusion need and sustained leukocytosis (≥13 x 10⁹/L). Mutations were also important in predicting leukemic transformation, notably DNMT3A, PTPN11 and ASXL1.