GATA2 Expression Level in Chronic Myeloid Leukemia (CML) Patients Correlates with Their Prognostic Scores and Is Associated with Disease Stage at Diagnosis

The Sokal, Hasford and EUTOS scoring systems have been successively proposed to predict the outcome of CML patients in chronic phase (CP-CML), both in terms of disease response and survival. The EUTOS score, which is the only one developed for patients treated with tyrosine kinase inhibitors, considers 2 risk groups based on spleen size and peripheral blood basophil percentage. No clear link has been identified between the covariates used to calculate those scores and CML molecular pathophysiology. To better delineate the molecular mechanisms underlying those scoring systems, a gene expression profiling study was performed on 48 CP-CML patients included in the Novartis ENEST1st study, for whom good quality peripheral blood total RNA was available at diagnosis. First, patients with low-risk (n=21) and high-risk (n=12) Sokal scores were compared, leading to identify 13 genes differentially expressed between those groups (q<.05). GATA2, which encodes a critical hematopoietic transcription factor, was among the 10 genes significantly overexpressed in high Sokal score CP-CMLs. Robust regression analyses confirmed that GATA2 expression level was correlated to the 3 prognostic scores, when considering those as continuous covariates (R²=.39, p<10^-4 [Sokal]; R²=.16, p=.007 [Hasford]; R²=.16, p=.007 [EUTOS]). Among the covariates used to calculate those scores, platelet count (R²=.35, p<10^-4), basophil (R²=.22, p=.0012), eosinophil (R²=.14, p=.01) and peripheral blood blast (R²=.13, p=.017) percentages were correlated to GATA2 expression level. Using a quantitative RT-PCR assay, GATA2 expression was subsequently assessed on an independent set of 80 CP-CML patients followed at Angers University Hospital from 2005 to 2014 (Sokal low, n=32; int, n=30; high, n=18 of which 7 EUTOS high). This validation set confirmed that GATA2 expression level was significantly correlated to the 3 prognostic scores (R²=.31, p<10^-4 [Sokal]; R²=.28, p<10^-4 [EUTOS]; R²=.11, p=.003 [Hasford]). Among the covariates used in those scores, basophil percentage (R²=.44, p<10^-4), platelet count (R²=.31, p<10^-4) and peripheral blood blast percentage (R²=.22, p<10^-4) were significantly correlated to GATA2 expression level in this second cohort. The ENEST1st gene expression dataset and a linear regression approach were then used to better define the role of GATA2 in CP-CMLs. This analysis identified 12 genes significantly correlated to GATA2 in terms of expression levels (q<10^-5). Half of those - CPA3, FCER1A, ENPP3, HDC, IL1RL1, SLC45A3 - were basophil-related genes. Interestingly, in normal umbilical cord blood-derived myelopoiesis, the expressions of GATA2, CPA3, FCER1A, and HDC were significantly correlated and coexpressed in common myeloid progenitors (CMP), granulocyte/monocyte progenitors (GMP) and megakaryocyte/erythroid progenitors (MEP) (GEO Microarray, GSE24759). Transcription factor ChIP-Seq data from the ENCODE project confirmed that, in K562 cell line, among the 12 genes identified above, GCSAML, EPAS1, FCER1A, HDC, IL1RL1, MGAT3, RDH11 and SLC45A3 were direct GATA2 targets. To extend this analysis, the gene expression dataset from Radich et al study (Proc Natl Acad Sci USA 2006;103:2794-9 - GEO Microarray, GSE4170), based on 108 peripheral blood samples from patients with CP-CML (n=57), accelerated phase (AP)-CML (n=17) or blastic crisis (BC)-CML (n=22 myeloid BC, n=12 B-cell lymphoblastic BC) were reanalyzed, and confirmed that GATA2 expression level was significantly correlated to the one of the basophil-related genes FCER1A, HDC and SLC45A3 in CP-CMLs. It also showed that GATA2 expression level significantly increased from CP-CMLs to myeloid BC-CMLs, whereas it was expressed at the lowest levels in lymphoblastic BC-CMLs. Similar results were observed when analyzing the dataset from Cramer-Morales et al study (Blood 2013;122:1293-304 - GEO Microarray, GSE47927), in which GATA2 was expressed at significantly higher levels in the hematopoietic stem cell, multipotent progenitors, CMP, GMP, and MEP compartments in myeloid BC-CMLs, when compared to the same compartment in AP-CMLs, CP-CMLs or in healthy volunteers. Therefore, GATA2 is a key gene affecting the outcome of CP-CMLs and the evolution into myeloid BC-CMLs. GATA2 is associated with a basophil-related gene expression signature that might explain the prognostic influence of the basophil percentage within the EUTOS score.