Abstract
Chronic myelomonocytic leukemia (CMML) is a clonal stem cell disorder with overlapping features between myelodysplastic syndromes and myeloproliferative neoplasms. Numerous models exist for CMML prognostication, with more recent studies suggesting (JCO 201; 31:2428) or refuting (Leukemia 2013; 27:1504) the prognostic contribution of ASXL1 mutations. Furthermore, SETBP1 mutations were recently shown to be associated with shortened overall survival (OS) in CMML (Leukemia 2013; 10:1038). In the current international study, we examine these issues in a larger cohort of 431 patients.
431 patients with WHO-defined CMML were included in the study. 235 (55%) were seen at the Mayo Clinic from 1997 through 2012. The remainders were from the French CMML registry (JCO 201; 31:2428). All patients underwent bone marrow (BM) examination and cytogenetic evaluation at diagnosis. DNA analysis for spliceosome component mutations (SRSF2, SF3B1 and U2AF1), ASXL1 and SETBP1 mutations were carried out on BM specimens obtained at diagnosis. In order to address the aforementioned discrepancy regarding the prognostic impact of ASXL1 mutations, relevant analyses in the Mayo cohort were first performed with and without inclusion of missense ASXL1 mutations. ASXL1 mutations from the French cohort did not include missense mutations. We evaluated the prognostic relevance of ASXL1 and SETBPI mutations, as well as several other clinical and laboratory parameters including those previously identified by the MDAPS (Blood 2002;99:840) the Spanish cytogenetic risk stratification (Haematologica 2011;96:375), and the Mayo prognostic model (Leukemia 2013;27;1504).
Among the 431 study patients, 286 (66%) were males and median age was 73 years (range, 17-93 years). There were 368 (85%) patients with CMML-1 and the remainder had CMML-2. At a median follow-up of 23 months, 260 (60%) deaths and 70 (16%) leukemic transformations were documented. Median survivals were 38 months for CMML-1 and 24 months for CMML-2 (p=0.11). Mutational frequencies were 44% (173/390) for SRSF2, 6% (23/379) for SF3B1, 7% (27/387) for U2AF1, 38% (164/411) for ASXL1 (excluding missense mutations), and 5% (21/431) for SETBP1. Risk stratification was, based on i) Mayo prognostic model: 172 (40%) high, 151 (35%) intermediate and 94 (25 %) low risk, ii) MDAPS: 15 (3%) high, 73 (17%) intermediate-2, 125 (29%) intermediate-1 and 218 (50%) low risk and iii) Spanish cytogenetic stratification system: 316 (73%) low, 43 (10%) intermediate and 50 (12%) high risk.
In the Mayo cohort, univariate analysis revealed that the exclusion of missense mutations changed the prognostic impact of ASXL1 mutations from non-significant (p=0.08) to significant (p=0.001). Accordingly, all subsequent analyses excluded missense ASXL1 mutations.
In univariate analysis, lower hemoglobin (p<0.0001), lower platelet count (p=0.0027), higher absolute monocyte count (AMC) (p<0.0001), higher absolute lymphocyte count (ALC) (p=0.0002), circulating immature myeloid cells (IMC) (P<0.0001), cytogenetic risk stratification (p<0.0001) and ASXL1 mutations (p<0.0001) were significant for OS. In multivariable analysis, lower hemoglobin (p=0.0001; RR 2, 99% CI 1.6-2.6), lower platelet count (p=0.002; RR 1.5, 99% CI 1.2-1.9), higher AMC (p=0.0002; RR 2.2, 99% CI 1.6-3.1) and ASXL1mutations (p=0.0009; RR 1.9, 99% CI 1.5-2.4) retained their independent negative prognostic impact.
Similarly, in univariate analysis, leukemia-free survival (LFS) was negatively affected by age (p=0.0015), lower hemoglobin (p=0.0002), lower platelet count (p=0.0002), higher AMC (p<0.0001), higher ALC (p=0.0001), circulating IMC (p<0.0001), BM blasts (p<0.0001), and cytogenetic risk stratification (p=.0002). ASXL1 (p=0.17) and SETBP1(p=0.87) mutations were not found to be significant. In multivariable analysis, lower platelet count (p=0.0005), higher AMC (P=0.0042), circulating IMC (P=0.008) and cytogenetic risk stratification (p=0.009) retained their independent negative prognostic impact.
In the current international study of a large cohort of patients with CMML, we confirm and clarify the independent prognostic relevance of ASXL1 mutations. The relatively high frequency of ASXL1 mutations in CMML warrants its inclusion in contemporary prognostic models.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.
